a flexible, modular approach to integrated space...
TRANSCRIPT
A FLEXIBLE, MODULAR APPROACH TO INTEGRATED SPACE EXPLORATION
CAMPAIGN LOGISTICS MODELING, SIMULATION, AND ANALYSIS
by
Paul Thomas Grogan
B.S. Engineering Mechanics and Astronautics
University of Wisconsin, 2008
Submitted to the Department of Aeronautics and Astronautics
in partial fulfillment of the requirements to the degree of
Master of Science in Aeronautics and Astronautics
at the
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
September 2010
© Massachusetts Institute of Technology.
All rights reserved.
Signature of Author…………………………………………………………………………………
Department of Aeronautics and Astronautics
July 30, 2010
Certified by…………………………………………………………………………………………
Olivier L. de Weck
Associate Professor of Aeronautics and Astronautics and
of Engineering Systems
Thesis Supervisor
Accepted by………………………………………………………………………………………...
Eytan H. Modiano
Associate Professor of Aeronautics and Astronautics
Chair, Committee on Graduate Students
2
A FLEXIBLE, MODULAR APPROACH TO INTEGRATED SPACE EXPLORATION
CAMPAIGN LOGISTICS MODELING, SIMULATION, AND ANALYSIS
by
Paul Thomas Grogan
Submitted to the Department of Aeronautics and Astronautics
on July 30, 2010, in partial fulfillment of the
requirements to the degree of
Master of Science in Aeronautics and Astronautics
3
Abstract
A space logistics modeling framework to support space exploration to remote environments is the target
of research within the MIT Space Logistics Project. This thesis presents a revised and expanded
framework providing capabilities to analyze a new set of explorations using a generalized resource flow
through a time-expanded network to satisfy exploration demands. The framework is both flexible to
model a wide range of destinations using mixed levels of fidelity and modular to enable future expansion
through interfaces.
The SpaceNet software tool implements the space logistics modeling framework, providing integrated
modeling and simulation capabilities for quantitative space exploration campaign analysis. Discrete event
simulation identifies logistical infeasibilities and provides quantitative measures of exploration
effectiveness to guide trade studies or other campaign analyses. SpaceNet 2.5, a Java executable with an
extensive graphical user interface, has been publicly released under an open source license.
Four case studies are presented as examples of the modeling framework applied to relevant exploration
campaigns. A resupply of the International Space Station from 2010-2015 includes 77 flights of seven
different vehicles from six launch sites to investigate the supply capacity under existing resupply
strategies. A near-Earth asteroid exploration details a two crew, 14-day tele-operated mission at 1999-
AO10 to establish the feasibility requirements of using modified Constellation vehicle architectures. A
lunar outpost exploration models the buildup of infrastructure and surface excursions leading to
continuous human presence over 21 missions and seven years. Finally, a Mars surface exploration models
the ten launches and in-space nuclear thermal rocket propulsion required to send a crew of four to the
surface of Mars for a 531-day exploration.
Finally, a usability experiment is presented to demonstrate the usability and efficiency of the SpaceNet
tool as compared to independent analysis methods. Seven test subjects were tested, five using SpaceNet
and two control subjects using spreadsheet-based methods, to analyze and establish the feasibility of a
near-Earth object mission. The median SpaceNet subject required 35 minutes to complete the analysis,
compared to a median of 113 minutes for the control subjects.
Thesis Supervisor: Olivier L. de Weck
Title: Associate Professor of Aeronautics and Astronautics and of Engineering Systems
4
(This Page Intentionally Blank)
5
Acknowledgements
There are many individuals in need of recognition for their substantial impact on this project. First, I
would like to thank Prof. Olivier de Weck for his advising and mentoring over the past two years. He has
generously supported both my academic and professional development by providing countless
opportunities to participate in projects, committees, conferences, and meetings, all of which have
dramatically supplemented my research.
I would be lost without the help of my family and closest friends. I would like to thank my father Tom for
encouraging me to always accept new challenges, my mother Kathy for showing what is truly important
in life, my sister Sarah for her selfless attitude, and my brother Brian for his loyal companionship. Special
recognition is reserved for Natasha Lewis, for whom frequent visits are the least I can offer for the
support she has provided.
A substantial portion of this project is built on the work of others. In particular, I would like to recognize
Nii Armar, Sarah Shull, and Afreen Siddiqi for their development contributions within the space logistics
modeling framework. Also, I would like to thank my colleagues at MIT for their collaboration and
assistance in case study development and usability testing. In particular, I would like to thank Ariane
Chepko, Torin Clark, Ben Corbin, Ivo Ferreira, Alessandro Golkar, Jennifer Green, Abe Grindle, Arthur
Guest, Tak Ishimatsu, Justin Kaderka, Greg O‟Neill, Basant Sagar, Narek Shougarian, Daniel Selva,
Anthony Wicht, and Howard Yue.
A portion my involvement was supported by NASA Strategic University Research Partnership (SURP)
contract number 1344341. Our contacts at JPL including Gene Lee, Elizabeth Jordan, and Dr. Robert
Shishko provided tremendous expertise and were instrumental in guiding the project in the right direction.
This thesis is dedicated to the memory of my grandfather, Prof. Paul J. Grogan (1918-2008). He served as
an inspiration from my earliest childhood and contributed greatly to my love of mathematics applied
through engineering.
6
(This Page Intentionally Blank)
7
Contents
1 Introduction ...................................................................................................................................... 14 1.1 Background ................................................................................................................................. 15
1.1.1 The Vision for Space Exploration ....................................................................................... 15 1.1.2 MIT Space Logistics Project ............................................................................................... 16 1.1.3 Human Space Flight Review ............................................................................................... 17
1.2 Project Motivation ...................................................................................................................... 17 1.2.1 Space Logistics “-ilities” ..................................................................................................... 18 1.2.2 Flexible Modeling Framework ............................................................................................ 18 1.2.3 Modular Implementation Architecture ................................................................................ 20 1.2.4 User Community ................................................................................................................. 22
1.3 Related Research and Literature ................................................................................................. 22 1.4 Project Overview ........................................................................................................................ 23
2 Domain Model .................................................................................................................................. 25 2.1 Network Components Model ...................................................................................................... 25
2.1.1 Nodes .................................................................................................................................. 25 2.1.2 Edges ................................................................................................................................... 27
2.2 Resource Model .......................................................................................................................... 28 2.3 Element Model ............................................................................................................................ 31
2.3.1 Element Hierarchy .............................................................................................................. 31 2.3.2 Element Parts ...................................................................................................................... 34 2.3.3 Element States ..................................................................................................................... 34 2.3.4 Element Demand Models .................................................................................................... 35
3 Campaign Modeling and Analysis .................................................................................................. 39 3.1 Network Model ........................................................................................................................... 39 3.2 Mission Model ............................................................................................................................ 40
3.2.1 Mission Demand Models .................................................................................................... 40 3.2.2 Mission Events .................................................................................................................... 40 3.2.3 Spatial Simulator ................................................................................................................. 43
3.3 Manifest Model ........................................................................................................................... 43 3.3.1 Demand Simulator .............................................................................................................. 44 3.3.2 Demand Aggregation .......................................................................................................... 48 3.3.3 Demand Packing ................................................................................................................. 48 3.3.4 Container Manifesting ......................................................................................................... 49
3.4 Campaign Analysis ..................................................................................................................... 49 3.4.1 Campaign Simulator ........................................................................................................... 49 3.4.2 Measures of Effectiveness .................................................................................................. 50
4 SpaceNet 2.5 Implementation .......................................................................................................... 51 4.1 Development and Release ........................................................................................................... 51
4.1.1 Development Methodology ................................................................................................. 52 4.1.2 Development Tools ............................................................................................................. 52 4.1.3 Public Release ..................................................................................................................... 53 4.1.4 Documentation .................................................................................................................... 53
4.2 Data Sources ............................................................................................................................... 54 4.2.1 Data Editor .......................................................................................................................... 55 4.2.2 Element Sizing Tool............................................................................................................ 56
4.3 Graphical User Interface ............................................................................................................. 56
8
4.3.1 Campaign Panel .................................................................................................................. 57 4.3.2 Network Tab ....................................................................................................................... 58 4.3.3 Missions Tab ....................................................................................................................... 58 4.3.4 Demands Tab ...................................................................................................................... 59 4.3.5 Manifest Tab ....................................................................................................................... 63 4.3.6 Simulation Tab .................................................................................................................... 65
5 SpaceNet 2.5 Applications ............................................................................................................... 68 5.1 Case Studies ................................................................................................................................ 68
5.1.1 International Space Station Resupply ................................................................................. 68 5.1.2 Near-Earth Object Sortie ..................................................................................................... 79 5.1.3 Lunar Outpost Buildup ........................................................................................................ 84 5.1.4 Mars Exploration................................................................................................................. 94
5.2 Usability Experimentation ........................................................................................................ 103 5.2.1 Testing Procedure ............................................................................................................. 103 5.2.2 Results and Discussion...................................................................................................... 105 5.2.3 Discussion ......................................................................................................................... 109
6 Conclusions ..................................................................................................................................... 110 6.1 Significant Contributions .......................................................................................................... 110 6.2 Future Work .............................................................................................................................. 110 6.3 Integrated Analysis Strategic Roadmap .................................................................................... 111
6.3.1 Analysis Scope .................................................................................................................. 112 6.3.2 Stochastic Analysis ........................................................................................................... 114
Appendix A Object Model Diagrams .............................................................................................. 117
Appendix B Optimal Manifesting Methods ................................................................................... 122 Modeling Cargo Manifests ................................................................................................................... 122 Campaign Feasibility and Manifest Optimization ................................................................................ 124
References ................................................................................................................................................ 126
9
List of Figures
Figure 1: MIT space logistics project timeline. The project has followed development and
changes to the NASA Human Space Flight program. .................................................................... 15 Figure 2: Use case scenario networks. (a) Lunar sortie. (b) ISS resupply post shuttle retirement.
(c) Lunar hub-spoke (outpost). (d) Lunar spoke-hub (global exploration). (e) Martian
human-robotic exploration. ........................................................................................................... 20 Figure 3: Procedural programming diagram. Procedures called by a monolithic GUI edit portions
of a shared memory. ....................................................................................................................... 21 Figure 4: Object-oriented programming diagram. Objects reside in memory and carry functional
capabilities limited in scope. Objects are accessed by parallel GUI components. ....................... 21 Figure 5: Lagrange point positions. L1, L2, and L3 are stable locations, L4 and L5 are unstable
and require station-keeping. Not to scale. ..................................................................................... 27 Figure 6: Surface edge example. Locations 1, 2, and 3 are connected with three surface edges. .............. 27 Figure 7: Space edge examples. Edges correspond to propulsive burn sequences. The burns are
highlighted for the space edge between nodes LEO and LLPO. .................................................... 28 Figure 8: Flight edge examples. Edges correspond to pre-determined flight architectures with
crew and cargo limits..................................................................................................................... 28 Figure 9: Element hierarchy. A hierarchy of elements provides specialized capabilities extensible
for future expansion while establishing a common set of inherited properties. ............................ 32 Figure 10: Part commonality with shared resource types. The Power Plant and Lander elements
share a common Power Supply resource. ...................................................................................... 34 Figure 11: Element reconfiguration sequence. An element may experience several operational
states throughout a campaign including dormant before delivery, active or quiescent
during operation, and decommissioned state to signal availability of parts for
scavenging. .................................................................................................................................... 35 Figure 12: Discrete event time effect on linear demands. Discrete time simulation aggregates
linear demands at each event execution. ....................................................................................... 36 Figure 13: Sparing by mass using a parts list. The sparing by mass demand model optionally can
access an element’s part list to generate unspecified, pressurized, and unpressurized
spares demands. ............................................................................................................................. 37 Figure 14: Campaign modeling and analysis. Campaigns are modeled in three parts: a network,
missions, and a manifest. When the campaign is both spatially feasible and logistically
feasible, analysis can provide visualizations and measures of effectiveness. ................................ 39 Figure 15: Network bat chart. A bat chart shows a time-expanded network and element movement
over time. ....................................................................................................................................... 39 Figure 16: Spatial simulation process. The spatial simulator sequentially executes events,
updating the network state and adding events when necessary, and may generate spatial
errors. ............................................................................................................................................ 43 Figure 17: Manifest model creation sequence. Manifesting requires the generation of demands,
aggregated to the location and time of demand. Resources are then packed into
containers which subsequently are manifested onto carriers for transport. ................................. 44 Figure 18: Demand simulation process. The demand simulator executes events, updating the
network state and adding events as necessary, may generate spatial errors, and logs
unsatisfied demands for analysis and manifesting. ........................................................................ 45 Figure 19: Impact of discretization policy. Sample demands for two elements with common
discrete demands. AP is the aggregation policy value. (a) No discretization treats the
demands as continuous. (b) Discretization by element aggregates items on a per-element
basis. (c) Discretization by location aggregates items on a per-location basis, enabling
spares pooling to locally decrease and smooth demands. ............................................................. 46
10
Figure 20: Impact of aggregation policy. The item aggregation policy determines at which point
demands are aggregated into whole units. (a) No discretization treats demands as
continuous. (b) A policy of 0.0 aggregates items at the first demand. (c) A policy of 0.5
aggregates items when half units are accumulated. (d) A policy of 1.0 aggregates items
after unit demands.......................................................................................................................... 47 Figure 21: Campaign transports and exploration periods. Each transportation arrival marks the
start of its corresponding exploration period. ............................................................................... 48 Figure 22: Demand packing. Aggregated demands are split into groups corresponding to how the
demands are packed within individual resource containers. ......................................................... 48 Figure 24: Campaign simulation process. The campaign simulator executes events, updating the
network (system) state and adding events as necessary, and logs measures of
effectiveness. .................................................................................................................................. 49 Figure 23: Multi-transport manifesting. Containers manifested on transport 3 (T3) are demanded
during exploration period 2 (E2) and must be supplied by either T1 or T2. ................................. 49 Figure 26: SpaceNet 2.5 development timeline. SpaceNet 2.5 R1 was released after initial
delivery to NASA JPL. SpaceNet 2.5 R2 is currently under development. .................................... 51 Figure 25: SpaceNet logo. ........................................................................................................................... 51 Figure 27: Early SpaceNet GUI prototype mockup. Prototype mockups assist development
iterations by walking through the campaign definition without time-intensive coding
investment. ..................................................................................................................................... 52 Figure 28: Data table relationships. Data is divided across several tables to represent parent-
child relationships between objects. .............................................................................................. 54 Figure 29: SpaceNet data editor user interface. The data editor provides a GUI for any data
source implementation. This example edits an edge from a spreadsheet database. ...................... 55 Figure 30: Element sizing model. An element sizing model creates element designs using
customized model input parameters. .............................................................................................. 56 Figure 31: SpaceNet 2.5 user interface. The SpaceNet GUI guides users through the process of
modeling and analyzing exploration campaigns. .......................................................................... 57 Figure 32: SpaceNet network visualization. Network for a lunar campaign including surface
(yellow) and orbital (red) nodes, and surface (green), space (red), and flight (yellow)
edges. ............................................................................................................................................. 58 Figure 33: Campaign visualizations. (a) The process bat chart displays mission events, processes,
and transports in time-expanded network. (b) The element bat chart displays individual
element movement in time-expanded network. .............................................................................. 59 Figure 34: Spatial simulation errors. Spatial simulation errors display (a) in the mission events
list and (b) in the mission event dialog to alert the user of error conditions. ................................ 59 Figure 35: Scenario feasibility visualization. Cumulative raw and remaining transport capacity
must always exceed estimated cumulative demands for feasible campaigns. ................................ 60 Figure 36: Supply network visualization. Line width represents net transport capacity and circle
diameter represents aggregated exploration period demand mass. .............................................. 61 Figure 37: Element demand history visualization. Plots demands for spares (COS 4) and packing
mass estimates (COS 5) generated by a habitat element. The non-linear response results
from reconfigurations of element state during crewed and un-crewed periods. ........................... 61 Figure 38: Commonality and scavenging analysis. Displays scavenged common parts filtered by
source element and type of resource. ............................................................................................. 62 Figure 39: Repair policy for optimal mass savings. The set of items to repair to achieve optimal
mass savings is found by sorting the repairable items in descending order by the benefit-
cost ratio. ....................................................................................................................................... 63 Figure 40: Packing and manifesting interface. The manifesting tab helps the user to create a valid
cargo manifest by guiding through the process of packing and manifesting containers. .............. 63
11
Figure 41: Network state history. The network state history shows the system state after each
event execution. Elements can be inspected for contents including remaining propellant
mass. .............................................................................................................................................. 66 Figure 42: Relative exploration capability history. REC does not increase monotonically over a
campaign due to non-exploration mass launched. ......................................................................... 66 Figure 43: Location resource history. The resource history during an example campaign shows a
build-up of COS 5, representing empty resource containers. ........................................................ 67 Figure 44: ISS resupply network. Visualization of Earth launch sites, landing zones, and the ISS
in orbit. ........................................................................................................................................... 69 Figure 45: ISS resupply mission bat chart. Green squares are element instantiations, yellow lines
are flight transports, black squares are elements removed from simulation. ................................ 75 Figure 46: ISS resupply scenario feasibility. Cumulative plot of raw supply vehicle capacity,
remaining capacity after pre-manifested elements, and aggregated demands indicating
feasibility. ....................................................................................................................................... 76 Figure 47: ISS resupply feasibility with Cygnus COTS failure. Without the Cygnus spacecraft,
steady-state infeasibilities fist start to appear in early 2013 and only worsen into 2014
and 2015. ....................................................................................................................................... 77 Figure 48: ISS resupply with advanced water recovery. An advanced water recovery system is
delivered in July 2013 reducing demands for water from 3.5 to 0.5 kg/person/day. ..................... 78 Figure 50: NEO sortie network. Visualization of space edges between launch, low-Earth orbit, the
near-Earth object 1999 AO10, and splashdown in the Pacific Ocean. ......................................... 79 Figure 49: Flexible path strategy. Timeline of milestones, destinations and capabilities of the
Flexible Path strategy, adapted from [21]..................................................................................... 79 Figure 51: NEO sortie infeasibilities. (a) The service module burn to return to Earth has a 526.8
m/s delta-v deficit without considering cargo. (b) The 111-day transit to 1999-AO10
alone exhausts the crew module cargo capacity. ........................................................................... 82 Figure 52: Modified NEO sortie service module resource history. The science instrument payload
(COS 6) is discarded. The service module propellant (COS 1) margin is 250 kg. ........................ 83 Figure 53: Modified NEO sortie crew module resource history. All crew consumables (COS 2)
are used in the crew module. Packing materials (COS 5) total about 100 kg of the cargo. .......... 83 Figure 54: Lunar surface exploration network. Visualization of flight edges to various surface
locations and surface edges between excursion sites near LSP. ................................................... 85 Figure 55: Lunar outpost process bat chart. Green squares are element instantiations, yellow lines
are flight transports, green lines are surface transports, orange and pink squares are
element movements and reconfigurations, blue lines are surface explorations, and black
squares are elements removed from simulation. ............................................................................ 91 Figure 56: Lunar outpost feasibility at LSP. Demands include science payloads, crew
consumables and spares with packing mass estimates included. .................................................. 91 Figure 57: Lunar outpost demands at LSP. COS 2: Crew Consumables, COS 3: Operational
Items, COS 4: Maintenance and Spares, COS 5: Packing Mass (Estimated), COS 6:
Science Payload, COS 7: Waste Disposal. .................................................................................... 92 Figure 58: Lunar outpost ISRU production and consumption. Mass of oxygen available for
consumption from (a) ISRU-1 and (b) ISRU-2. ............................................................................. 92 Figure 59: Lunar outpost measures of effectiveness. (a) Crew surface days across all surface
sites. (b) Exploration capability (crew-kg-days) across all surface sites. ..................................... 93 Figure 60: Mars exploration network. Visualization of space edges utilizing propulsive burns to
reach potential Martian surface sites. ........................................................................................... 94 Figure 61: Mars exploration process bat chart. Green squares are element instantiations, red lines
are space transports, yellow lines are flight transports, orange squares are element
movements and reconfigurations, and the blue line is a surface exploration. ............................... 99
12
Figure 62: Baseline Mars exploration crew demands. Under baseline crew member demand rates
with a crew of six, demands total 8.1 tons during transport to Mars, 20.8 tons during
surface operations, and 9.3 tons during return transport. ........................................................... 100 Figure 63: Modified Mars exploration crew demands. Using a 95% water closure loop and
reusable hygiene and waste disposal items, demands are be reduced to 3.6 tons during
transport to Mars, 7.6 tons during surface operations, and 4.1 tons during return
transport. ...................................................................................................................................... 100 Figure 64: Modified Mars exploration transport capacity. An additional CFC is included to satisfy
demands during transport and the cargo capacity for the MDAV and SHAB have been
increased to satisfy demands during surface operations. ............................................................ 102 Figure 65: User experiment baseline spacecraft. (a) Modified Ares V launch vehicle. (b) Modified
Orion crew exploration vehicle. .................................................................................................. 105 Figure 66: User experiment results. (a) SpaceNet subject task milestones. (b) Comparison of
treatment (SpaceNet) and control (spreadsheet) subjects’ total task time. Whiskers show
minimum and maximum recorded values..................................................................................... 108 Figure 67: Manifesting problem diagram. Optimization of the manifesting problem produces a
cargo manifest to support exploration goals during simulation. ................................................. 113 Figure 68: Scheduling problem diagram. Optimization of the scheduling problem produces a
mission sequence and cargo manifest to support exploration goals during simulation. ............. 113 Figure 69: Architecting problem diagram. Optimization of the architecture problem produces
mission and element architectures, a mission sequence, and a cargo manifest to support
exploration goals during simulation. ........................................................................................... 114 Figure 70: Deterministic versus stochastic demands. (a) Deterministic demands provide a single
campaign valuation. (b) Stochastic demands require stochastic manifesting and result in
a distribution of campaign valuations. ........................................................................................ 115 Figure 71: Stochastic event simulation. The manifesting process must be continuously revised
during simulation to react to resolved uncertainties. .................................................................. 116 Figure 72: Example object model diagram. The Vehicle interface has two subclass objects, Car
and Airplane. A car is comprised of an engine object and aggregates Passenger objects. ........ 117 Figure 73: Location objects. Locations aggregate Elements. The Location interface is expanded
by the Node and Edge interface, which both have several subclasses. ........................................ 118 Figure 74: Resource objects. Three subclasses of the Resource interface include Discrete,
Generic, and Continuous Resources. ........................................................................................... 118 Figure 75: Element objects. The Element interface is comprised of States and Parts. States are
comprised of Element Demand Models and Parts aggregate Resources. ................................... 118 Figure 76: Element demand model objects. Element Demand Models aggregate Resources.
Subclasses include Linear and Sparing by Mass Demand Models. ............................................. 119 Figure 77: Element object hierarchy. All elements inherit attributes from the base Element class.
Subclasses include Human Agents, Resource Containers, and Carriers. Resource Tanks
are subclasses of Resource Containers and Surface and Propulsive Vehicles are
subclasses of Carriers. ................................................................................................................. 119 Figure 78: Campaign objects. A campaign model is comprised of a Network, Mission, and
Manifest models. .......................................................................................................................... 120 Figure 79: Mission demand model objects. Mission demand models are very similar to element
demand models in that they aggregate Resources. ...................................................................... 120 Figure 80: Core event objects. Core element-based events include creation, movement, removal,
and reconfiguration. Core resource-based events include addition, demand, and transfer. ....... 120 Figure 81: Composite event objects. Composite events use sequences of core events to build up to
complex functionality. .................................................................................................................. 121
13
List of Tables
Table 1: SpaceNet development history. .................................................................................................... 16 Table 2: Development use cases summary. ................................................................................................ 19 Table 3: Surface node examples. ................................................................................................................ 26 Table 4: Orbital node examples. ................................................................................................................. 26 Table 5: Lagrange node examples. ............................................................................................................. 27 Table 6: Common resource classes of supply. ............................................................................................ 29 Table 7: Continuous resource examples. .................................................................................................... 30 Table 8: Discrete resource examples. ......................................................................................................... 30 Table 9: Common element classes of supply. ............................................................................................. 31 Table 10: Resource container examples. ..................................................................................................... 33 Table 11: Resource tank examples.............................................................................................................. 33 Table 12: Annual demands using a sparing by mass model with a parts list. ............................................. 38 Table 13: Crew consumables mission demand model parameters. ............................................................. 40 Table 14: Element-based core events list. ................................................................................................... 41 Table 15: Resource-based core events list. ................................................................................................. 41 Table 16: Composite events list. ................................................................................................................. 42 Table 17: Campaign measures of effectiveness. ......................................................................................... 50 Table 18: Campaign options list. ................................................................................................................ 57 Table 19: ISS resupply nodes...................................................................................................................... 69 Table 20: ISS resupply edges. ..................................................................................................................... 70 Table 21: ISS resupply elements. ................................................................................................................ 71 Table 22: ISS resupply missions 0-22 (2010-2011). ................................................................................... 72 Table 23: ISS resupply missions 23-45 (2011-2013). ................................................................................. 73 Table 24: 2014 ISS resupply missions 46-68 (2013-2015). ........................................................................ 74 Table 25: ISS resupply missions 69-77 (2015). .......................................................................................... 75 Table 26: NEO sortie nodes. ....................................................................................................................... 80 Table 27: NEO sortie edges. ....................................................................................................................... 80 Table 28: NEO sortie elements. .................................................................................................................. 81 Table 29: NEO sortie mission events. ......................................................................................................... 81 Table 30: Lunar outpost nodes. ................................................................................................................... 86 Table 31: Lunar outpost edges. ................................................................................................................... 86 Table 32: Lunar outpost primary elements. ................................................................................................ 87 Table 33: Lunar outpost secondary elements. ............................................................................................. 88 Table 34: Lunar outpost missions 1-10. ...................................................................................................... 89 Table 35: Lunar outpost missions 11-21. .................................................................................................... 90 Table 36: Mars exploration nodes. .............................................................................................................. 95 Table 37: Mars exploration edges ............................................................................................................... 95 Table 38: Mars exploration elements .......................................................................................................... 96 Table 39: Mars exploration mission events. ............................................................................................... 98 Table 40: Nuclear thermal rocket specific impulse sensitivity. .................................................................. 99 Table 41: User experiment subject comparison. ....................................................................................... 103 Table 42: Propulsive fuel margins. .......................................................................................................... 106 Table 43: Post-test questionnaire results. .................................................................................................. 107 Table 44: User experiment feasible designs. ............................................................................................ 108 Table 45: Levels of space exploration analysis......................................................................................... 112
14
1 Introduction
As humans grasp the reigns of technology and ride beyond our planet‟s tenuous atmosphere to more
distant destinations over extended time periods, the importance of strategic planning becomes paramount
to ensuring the safety and success of future space exploration missions. Over the next 50 years, mission
architectures are expected to transition from single, independent sorties to tightly-integrated campaigns
spanning many years and involving several stakeholder organizations. As a system-of-systems, a space
exploration campaign will require sophisticated logistics and supply chain planning to maintain human
presence in remote, hostile environments.
Following the announcement of a new era of human spaceflight and exploration in 2004, the American
Institute of Aeronautics and Astronautics (AIAA) reorganized the Space Logistics Technical Committee,
defining space logistics as [1]:
The theory and practice of driving space system design for operability, and of managing the flow
of materiel, services, and information needed throughout the space system lifecycle.
Addressing the logistics of space exploration is challenging due to two major differences from terrestrial
analogs. First, the physics of rocket propulsion provide only a minute fraction of launch mass (typically
well below 1%) for resources and items needed during exploration. This narrow margin forces careful
selection of what cargo to bring and makes multi-level packing and packaging a high priority. Second, the
dynamics of orbital trajectories significantly constrains transportation schedule and duration. If a critical
item fails during an exploration, it may take weeks or months to deliver a replacement with no
alternatives for resupply.
The first grand experiment in space logistics is already underway. Over the next ten years, the completed
International Space Station will serve as the exploration frontier in low Earth orbit, receiving up to eight
different vehicles from six launch sites to support a crew of six and over 350 tons of infrastructure. Future
campaigns to return to the Lunar or Martian surface will require even more substantial planning to
manage long-duration transportation, time-dependent launch windows, and high radiation environments
while providing robustness to inevitable failures and maintaining high exploration capability.
The goal of this thesis is to expand on an existing modeling framework developed under the MIT Space
Logistics Project to address the challenges of space logistics to evaluate exploration feasibility and
quantify exploration value for new and interesting space exploration campaigns.
15
1.1 Background
The efforts to model, simulate, and analyze the logistics behind space exploration campaigns are closely
tied with NASA‟s human space flight program as well as existing and continuing research areas at MIT.
The following sections describe the background research and human spaceflight events leading up to this
project, summarized in Figure 1.
Figure 1: MIT space logistics project timeline. The project has followed development and
changes to the NASA Human Space Flight program.
1.1.1 The Vision for Space Exploration
A new era of space exploration was announced by President George W. Bush in January 2004, putting the
United States on the path to long-term human and robotic exploration including a return to the Moon‟s
surface in preparation for a human exploration to Mars [2]. To achieve the goals, a tentative timeline was
set to develop a crew exploration vehicle by 2008, conduct the first human spaceflight mission by 2014,
and explore the Moon with robotic missions by 2008 and with crewed missions by 2020 [3].
Early architectural studies performed by NASA shaped the agency‟s plan of action to achieve the
Vision‟s goals. The Exploration Systems Architecture Study (ESAS), completed in 2005, developed the
Constellation Program, which included a general mission architecture of two launch vehicles: one
human-rated vehicle to carry crew (Ares I), and one heavy-lift vehicle (Ares V) to carry in-space vehicles,
infrastructure, and cargo [4]. The Lunar Architecture Team (LAT) Phase I, completed in 2006, developed
on a Global Exploration Strategy promoting multi-national cooperation for lunar explorations of
increasing duration, focusing on utilizing lunar in-situ resources to develop technology necessary for a
sustainable Mars mission [5].
16
1.1.2 MIT Space Logistics Project
In coordination with NASA‟s architectural studies, MIT started the Space Logistics Project to build a
research base supporting interplanetary supply chain management and logistical analysis necessary for
extended exploration campaigns. The project initially studied several terrestrial analogs to space
exploration, including operations in remote terrestrial environments such as the Arctic and Antarctic,
commercial supply chains, and military logistics operations, culminating in the development of a space
logistics framework [6, 7].
SpaceNet, a software tool implementing the space logistics framework, supports campaign analyses and
trade studies. SpaceNet models space exploration from a supply chain and logistics perspective, and has
been under continuous development over the past five years, as summarized in Table 1. As of 2008,
SpaceNet was a fully functional model resulting from two years of development and application within
the NASA community. Its analytical capabilities were demonstrated in the Constellation Program (CxP)
Integrated Design Analysis Cycle 2 (IDAC-2) in 2006, where SpaceNet was used to trade launch
architectures and propellant types and quantify performance drivers for the lunar campaign [8,9].
Table 1: SpaceNet development history.
Designation Timeframe Comments
SpaceNet 1.1 2005-2006 Prototype
SpaceNet 1.2 2006 Established visualizations and database
SpaceNet 1.3 2006-2007 Public release, scenarios focused on lunar sorties
SpaceNet 1.4 2007-2008 Scenarios focused on lunar campaigns, demand modeling
SpaceNet 2.0 2008 Internal prototype, code migration to Java, advanced
visualizations, SQL database
SpaceNet 1.3 was released to the public in 2007 as a MATLAB® application and graphical user interface
(GUI) supported by an Excel database [10]. The overall framework of space logistics campaign analysis
was further expanded upon and summarized in Shull‟s conference paper and thesis, establishing a solid
foundation for future work efforts [11, 12]. SpaceNet 1.4 included additional development to improve the
ability to analyze long-duration lunar surface campaigns in support of CxP IDAC-3 [13, 14].
In 2008, there was an effort to migrate SpaceNet from MATLAB® to a cross-platform, web-accessible
implementation. SpaceNet 2.0 served as an internal Java Web Start prototype that utilized modular
aspects of object-oriented programming to provide a platform for future extension and development. The
flexibility of the revised architecture was demonstrated by Armar‟s paper on cargo revenue management
techniques for determining optimal manifests [15].
17
1.1.3 Human Space Flight Review
Under increasing pressure to reconcile the forward budget for the Constellation Program and the Vision
for Space Exploration, the Obama Administration requested an independent review of NASA‟s Human
Space Flight (HSF) program in May 2009 [16]. The major concern was the federal budget allocated to
NASA could not support both the sustainment of the International Space Station and the development of
the new systems required to meet the program of record schedule. NASA established a blue-ribbon panel
of experts, chaired by Norman Augustine, to perform the review over the summer of 2009 [17].
The “Augustine Commission” submitted a final report in October 2009 outlining several options for the
future of the human space flight program [18]. The committee provided three classes of options: a
“constrained” option adhering to the forward budget profile but would not return to the moon before
2030, a “moon-first” option using an expanded budget to support a campaign similar to the program of
record with a lunar return in the mid-2020‟s, and a “flexible path” option using an expanded budget but
would first focusing on non-lunar missions, such as near-Earth objects (NEOs).
In February 2010, President Obama released the 2011 fiscal year budget, influenced by the findings of the
Augustine Committee [19]. Major decisions included extending funding for the International Space
Station though 2020, cancelling the Constellation Program, and renewing focus on commercial space
transportation. Overall, NASA‟s budget was to increase by $6.0 billion over five years, bringing the total
funding for NASA to $100 billion. Administrator Bolden was quoted that the new investments “…will
enable our path beyond low Earth orbit through development of new launch and space transportation
technologies, nimble construction capabilities on orbit, and new operations capabilities…” and may
include “…new and novel approaches to spaceflight such as in-orbit fuel depots and rendezvous and
docking technologies, and closed-loop life support systems…” [20]. However, as of the time of writing,
no final decisions have been approved by Congress as to the future of the HSF program.
1.2 Project Motivation
The space logistics framework and code used in SpaceNet resulted from several years‟ worth of iterative
effort. Also, the targeted analysis focus had shifted over time, from investigating individual sorties (single
missions) to long-duration campaigns and from lunar-specific campaigns to more general exploration
scenarios. Several areas were identified to motivate improvements to the existing model to progress
research and accommodate the programmatic changes within the HSF program.
18
1.2.1 Space Logistics “-ilities”
One of the topics of interest emerging from the CxP IDAC-3 analysis was the ability to model and
“quantify the „-ilities‟ as part of an end-to-end campaign scenario, such as the reconfigurability,
reusability, commonality, and repairability of elements” [13]. Representing emergent lifecycle properties
of the space exploration system, the “-ilities” could contribute to decisions in the early design phase of the
system elements. This addition was developed under a Strategic University Research Partnership (SURP)
with Caltech‟s Jet Propulsion Laboratory.
Implementation of these “-ilities,” however, extended beyond the existing domain of the space logistics
framework. To account for the lower-level nature of the analysis, the task was separated into two
components. The first would model the “-ilities” at the subsystem level; the second would propagate their
impacts back to the system and campaign level. The following features were targeted in SpaceNet to
handle this new domain of analysis:
1. Model the reconfigurability of systems such that duty cycles, functional capabilities, and operational
states can be chosen dynamically depending on the mission plan. Challenges include sufficiently
describing operational states and dynamically changing the underlying models generating demands
during simulation.
2. Analyze the reuse of systems beyond the utilization of static, pre-deployed assets, specifically surface
mobility systems. Challenges include decoupling elements from missions and persisting elements and
their effects on demands throughout the simulation.
3. Analyze the impact of commonality and scavenging between systems such as surface rovers,
habitation modules, and propulsive stages. Challenges include adding the ability to scavenge
components from existing elements.
4. Model surface repair activities and how they impact system availability, spare parts and tool
demands, and the time available for exploration. Challenges include adding the ability to choose
repair actions and evaluating the impact on the overall scenario.
1.2.2 Flexible Modeling Framework
Several “use case” scenarios were identified during the development phase to develop a flexible modeling
framework capable of adapting to new scenarios including those emerging from future programmatic
changes. Five cases of increasing difficulty, summarized in Table 2, were created to guide development
and prevent hard-coding assumptions.
19
Table 2: Development use cases summary.
Use Case Primary System Flights New Challenges
Lunar Sortie Earth-Moon 1 Match existing modeling capability
ISS Resupply 2010-2015 Earth 40 Non-zero initial state
Complex models may be required for
realistic demands
Lunar Hub-Spoke (Outpost) Earth-Moon 20 Long duration, detailed campaign
High level of element reuse
Dependence on ISRU technology
Lunar Spoke-Hub (Global) Earth-Moon 20 Surface-to-surface transportation
Martian Human-Robotic Earth-Mars 16 Infrequent launch windows
Variable flight durations
Consideration of low-thrust propulsion
Long-term crew health concerns
The Lunar Sortie models a single Constellation-class mission for a seven-day duration exploration of the
Lunar South Pole (LSP), matching existing modeling capability.
International Space Station (ISS) Resupply for 2010-2015 models the sustainment of ISS after shuttle
retirement utilizing a combination of International Partners (IP) and Commercial Orbital Transportation
Services (COTS).
Lunar Hub-Spoke (Outpost) models the build-up and sustainment of an outpost located on the rim of
Shackleton Crater at the Lunar South Pole. The surface infrastructure and mobility elements are delivered
over the span of seven years, leading up to a continuous human presence on the moon. The campaign is
influenced by mission architectures investigated by the Constellation Architecture Team (CxAT).
Lunar Spoke-Hub (Global Exploration) relies on reusable surface mobility elements to model a global
exploration of the moon over a seven year period. Initial missions are targeted at building up
infrastructure at the Lunar South Pole. Later missions perform portions of a long-term traversal to the
North Pole using mobile habitation elements.
The Martian use case models a precursor robotic exploration of the Martian surface, followed by human-
crewed missions based on Mars Design Reference Architecture 5.0 [21]. The robotic precursors are sent
to select one of three target surface locations for the human missions.
20
a)
b)
c)
d)
e)
Figure 2: Use case scenario networks. (a) Lunar sortie. (b) ISS resupply post shuttle
retirement. (c) Lunar hub-spoke (outpost). (d) Lunar spoke-hub (global exploration). (e)
Martian human-robotic exploration.
1.2.3 Modular Implementation Architecture
Aside from a flexible modeling framework, it was also a priority to design the tool in a modular fashion
to provide a platform from which to establish future extensions and development. The development of
SpaceNet up to version 1.4 used procedural programming methods in MATLAB®, common for
engineering computation. The underlying structure of procedural programs resembles a bus for both
memory and the user interface, shown in Figure 3. Data is managed in memory accessible to any function
called by the user via a monolithic user interface.
In typical development progression, the organization of memory often does not keep up with the creation
of procedures. Shortcuts enabling access to previously unrelated data contribute to a code base that is
difficult to understand and more challenging to track, as potentially every procedure could alter shared
21
memory. In addition, there may be limited documentation or definition of interfaces between procedures,
complicating extensions or future development.
Figure 3: Procedural programming diagram. Procedures called by a monolithic GUI edit
portions of a shared memory.
SpaceNet 2.0 took the first steps into the development of an object-oriented architecture. Object-oriented
methods establish object classes to couple memory structure and function and separate GUI functionality,
as shown in Figure 4. Memory is not widely accessible; rather it is limited to the scope of each object so
unrelated objects cannot alter each other‟s data, equivalently, no knowledge of unrelated objects is
necessary to understand an object. Interfaces between objects are also inherently defined, helping to
insulate against change propagation and enabling easier extensions.
Figure 4: Object-oriented programming diagram. Objects reside in memory and carry
functional capabilities limited in scope. Objects are accessed by parallel GUI components.
The graphical user interface (GUI) can be divided into components to mirror the object structure
providing specialized interfaces to individual objects. As the GUI is separated from the underlying
modeling code, changes and updates can be easily implemented without a substantial update or revision to
22
the model. This is especially important when considering serialized data (saved files), which often
requires identical models for compatibility but has no restriction on modifications to the GUI.
1.2.4 User Community
Another goal motivating additional SpaceNet development was the desire to open the MIT Space
Logistics Project to a wide user community, potentially extending to applications outside field of space
exploration logistics. A three-fold strategy was undertaken towards this goal.
First, the usability of the software tool was targeted for dramatic improvement. From the user‟s
perspective, user-centric and task-centric methods should be embraced to improve the user interface.
From the developer‟s perspective, a modular architecture should be developed, isolating and abstracting
components likely to be expanded upon in the future. Second, the software tool should be migrated to a
license-free platform as many potential users do not have access to a MATLAB® license. Finally, an
open source coding strategy was desired to provide transparency of the space logistics framework and
allow the user community to self-sponsor modifications or expansions as necessary.
1.3 Related Research and Literature
Aside from the previous references within the MIT Space Logistics Project, several research papers
investigate topics related to space exploration campaign modeling and simulation.
Cirillo provides an overview of a strategic analysis methodology that “… provides integrated analysis of
system performance over the full system life cycle…” [22]. Strategic analysis investigates three main
components of space exploration scenarios: performance, affordability, and risk. He identifies the
differences between macro-logistics (resource movement between locations) and micro-logistics
(resource movement within a location) and the interplay between deterministic and probabilistic analysis.
The methodology is demonstrated with an example scenario focusing on lunar surface exploration.
Stromgren takes a closer look at micro-logistics analysis by modeling gas/liquid and solid logistics usage
using system dynamics for a lunar outpost exploration scenario [23]. Pressurized Logistics Modules
(PLMs) and Cargo Transfer Bags (CTBs) are used as logistics carriers. A visualization module also
enables a graphical depiction of the consumables during simulation.
Andraschko presents Campaign Analysis Manifest Tool (CMAT), a deterministic software model to
support strategic analysis activities within the Constellation Architecture Team – Lunar (CxAT-Lunar)
[24]. CMAT is very similar to SpaceNet in that exploration scenario definitions are used as inputs and
23
logistics feasibility is evaluated via an iterative manifesting process. Logistics considered include
pressurized crew logistics (food, hygiene, clothing, operational supplies, etc.), pressurized and
unpressurized logistics and maintenance, pressurized and unpressurized science, and consumables
(oxygen, nitrogen, and water). After the CxP IDAC-3 study in which SpaceNet 1.4 was used, CMAT was
selected as the primary tool for future analysis of lunar architectures.
Within the context of space logistics “-ilities,” Siddiqi investigates the spare parts requirements for
missions utilizing reconfigurability and commonality between elements [25]. She describes the benefits of
commonality for scavenging spares from non-operational elements and presents a method for determining
spare parts demands for reconfigurable states where the element may transition between several
operational states with different demand profiles. The theory is demonstrated with the discrete event
simulation of an example Mars exploration mission.
Kline describes a hybrid parametric-analytic sparing model to estimate the spare parts needed to achieve
specified system availability [26]. Commonality of spare parts is investigated both within elements and
across elements. The sparing model is implemented in the Spacecraft Sustainability Model™ (SSM) and
demonstrated with an example Mars exploration mission based on a NASA design reference mission.
1.4 Project Overview
A decision was made to restructure the space logistics modeling framework from the ground up to
achieve the project goals. The modeling framework to represent space exploration scenarios is defined
using object-oriented methods. The lowest-level domain objects including locations, resources, and
elements at the core of any logistical analysis regardless of application are discussed in Chapter 2. The
higher-level campaign objects more specific to space exploration including missions, events, and
manifests and the discrete event simulation engines used to drive analysis are discussed in Chapter 3.
Parallel to the efforts in developing modeling frameworks, logic flow diagrams and modular prototypes
developed from a user‟s perspective guide how a scenario should be constructed and analyzed. A user
interface was generated by breaking down an exploration campaign into the high-level modules a user
would define, including the network definition, mission and transportation specification, demand
modeling, cargo manifesting, and simulation and visualization. The resulting software implementation,
SpaceNet 2.5, is discussed in Chapter 4.
Applications of SpaceNet are presented in Chapter 5. Four case studies illustrating the flexibility to model
vastly different scenarios are presented including initial results and analysis. The case studies include the
24
resupply of the International Space Station from 2010-2015, a sortie mission to a near-Earth object for
human tele-operated exploration, the buildup of a Lunar outpost leading to continuous human presence,
and a crewed surface exploration on Mars. In addition, a controlled user experiment is presented to
highlight potential benefits of using the SpaceNet tool over independent analysis methods. Finally,
Chapter 6 provides some concluding thoughts, including a summary of the primary contributions of this
project and a strategic outline towards expanding the space logistics framework in future research.
25
2 Domain Model
The core of the space logistics modeling framework is focused on the domain, or physical state of
components within an exploration campaign. Components of the domain have evolved development
within the MIT Space Logistics Project. A typical iteration loop starts with specialized capabilities later
expanded to more general cases through abstraction. For example, SpaceNet 1.4 generalized the SpaceNet
1.3 model to allow multi-mission campaigns. As a substantial change from previous models, however,
object-oriented methods are used to emphasize the concepts of modularity and flexibility. Object model
diagrams illustrating the modeling framework visually are available in Appendix A.
The domain model is divided into three categories: network components, resources, and elements.
Although specific to space logistics, the domain representation of the modeling framework is extensible
to new applications because of its focus on physical representation. In other words, any model for
logistics, exploration, or related applications will incorporate similar domain objects.
2.1 Network Components Model
Network components represent locations and paths containing elements and nested resources within a
time-expanded network. Using network terminology, nodes and edges are two types of network
components that serve as interfaces to specialized components. New implementations of an interface
would be interchangeable with the existing objects in any generic application, such as those in most
underlying simulation code.
2.1.1 Nodes
Nodes represent time-invariant locations in the campaign. For some nodes, such as surface locations on a
planetary body, this definition is clear as the surface location does not move over time. In other nodes,
such as stable elliptical orbits, the time-invariant nodal representation corresponds to the entire orbit.
The primary purpose of nodes is to set the limits of resource sharing. Any resources at a node are assumed
to be equally accessible to all co-located elements to satisfy demands. This definition means that
resources are shared at the level of nodes, or equivalently, demands for resources are aggregated to nodes.
In campaigns more focused on micro-logistics, this assumption may not be desired – for example,
resources may not be explicitly shared between space station nodes, or habitats at a surface outpost may
require independent resources to satisfy demands. In these cases, pseudo-nodes can be defined to handle
the desired level of sharing or demand aggregation even though the nodes may be in close proximity.
26
Subclasses of nodes include surface nodes corresponding to locations on a planetary body, orbital nodes
corresponding to stable orbits, and Lagrange nodes corresponding to Lagrange points between celestial
bodies. Nodes of any subclass may also be designated as source nodes, meaning they may act as a source
of resources. Launch sites such as the Kennedy Space Center (KSC) are most often designated as source
nodes. Source nodes play an important part in the manifesting process, detailed in Section 3.3.
Surface Nodes
Surface nodes represent a location on a planetary body given by a latitude and longitude. The latitude and
longitude values are primarily used to drive projections of the surface node onto the planetary body in
visualizations, but future development may utilize this information to support other calculations. Table 3
lists commonly-used surface nodes.
Table 3: Surface node examples.
Surface Node Abbrev. Body Latitude Longitude
Kennedy Space Center KSC Earth 28.6°N 80.6°W
Lunar South Pole LSP Moon 89.9°S 0.0°E
Apollo 17 Landing Site A17 Moon 20.2°N 30.8°E
Viking 1 Landing Site VK1 Mars 22.5°N 50.0°W
Orbital Nodes
Orbital nodes represent elliptical orbits about a celestial body given by an inclination and altitude
apoapsis and periapsis. As orbital nodes represent time-invariant locations, two elements co-located at an
orbital node are assumed to be co-positioned in orbit, though multiple pseudo-nodes could be created with
similar orbital parameters to enforce differentiation. Similar to surface nodes, the orbital parameters
presently drive visualizations rather than serving as inputs to calculations or logical instructions. Table 4
provides examples of commonly-used orbital nodes.
Table 4: Orbital node examples.
Orbital Node Abbrev. Body Inclination Apoapsis
[km]
Periapsis
[km]
Low Earth Orbit LEO Earth 28.5° 296 296
International Space Station ISS Earth 51.6° 360 347
Low Lunar Polar Orbit LLPO Moon 90.0° 100 100
Low Mars Equatorial Orbit LMEO Mars 0.0° 500 500
27
Lagrange Nodes
Lagrange nodes represent the five libration points
(zones of no net gravitational acceleration from the
celestial bodies), as illustrated for reference in Figure
5. Though current use of Lagrange points for logistics
is limited to observational satellites, future missions
may use them to maintain fuel or other supply depots.
Table 5 provides examples of commonly-used
Lagrange nodes.
Table 5: Lagrange node examples.
Lagrange Node Abbrev. Major
Body
Minor
Body Position
Sun-Earth L1 SEL1 Sun Earth 1
Sun-Earth L2 SEL2 Sun Earth 2
Earth-Moon L2 EML2 Earth Moon 2
2.1.2 Edges
Edges represent time-invariant connections between nodes. During simulation, elements traverse edges
using a transport event (see Section 3.2.2). Although an edge is a time-invariant path, the transport event
may impose time-dependent characteristics. Elements co-located on an edge are assumed to share
resources in the same way that elements co-located at a node share resources.
Subclasses of edges include surface edges corresponding to paths between surface nodes, space edges
corresponding to impulsive propulsion trajectories, and flight edges corresponding to known
transportation architectures simplified to eliminate
propulsive requirements.
Surface Edges
Surface edges represent paths between two surface
nodes, parameterized by a distance and traversed using
a surface transport. Surface edges are not directional,
meaning traversal is possible in both directions. Surface
edges enable movement of previously-deployed
elements to new locations while modeling demands
Figure 5: Lagrange point positions. L1, L2, and
L3 are stable locations, L4 and L5 are unstable
and require station-keeping. Not to scale.
Figure 6: Surface edge example. Locations 1,
2, and 3 are connected with three surface edges.
28
during transport. Figure 6 illustrates surface edges between three surface nodes distances in kilometers.
Space Edges
Space edges represent directional paths between two nodes
requiring a series of propulsive burns for traversal. Each
burn is designated by an execution time relative to the start
of the transport, required change in velocity (delta-v), and a
specification of using either a propulsive vehicle Orbital
Maneuvering System (OMS) or Reaction Control System
(RCS). Figure 7 illustrates space edges used in lunar
transportation network to represent launch, trans-lunar
injection/lunar orbit insertion, descent, ascent, and trans-
Earth injection.
Flight Edges
Flight edges represent an abstracted path between two nodes
without relying on propulsive burns. They commonly use a
standard flight architecture known to feasibly deliver a set
amount of cargo and crew (e.g. a launch vehicle). Flight
edges are used when detailed analysis of individual space
edges and urns is not needed; common applications include
extended-duration campaigns with many missions utilizing
similar transportation methods between common nodes. A
flight is described by its duration, the crew capacity, and the
cargo capacity. Figure 8 illustrates an example network using
flight edges to abstract a series of space transports.
2.2 Resource Model
Resources are substances consumed to satisfy demands. Resources only specify types of substances rather
than the instantiated items expended during simulation – the name “resource type” has been used in
previous modeling frameworks to make this distinction. Resources must always be contained by
elements, specifically resource containers, which maintain the quantity of each resource.
Figure 7: Space edge examples. Edges
correspond to propulsive burn sequences.
The burns are highlighted for the space
edge between nodes LEO and LLPO.
Figure 8: Flight edge examples. Edges
correspond to pre-determined flight
architectures with crew and cargo limits.
29
All resources are assigned a functional class of supply (COS) based on military and NASA techniques for
classifying cargo by its function [27]. Classes of supply are used to abstract and group similar resources to
simplify demand models and visualizations. There are ten primary classes of supply and many sub-classes
of supply that define cargo in more detailed terms. As a general guideline, classes 1-4 and 7 are
commonly used for resources and 5, 6, 8, 9, and 10 for elements. Table 6 lists the current set of classes of
supply commonly used for resource classification.
Table 6: Common resource classes of supply.
COS Description and Sub-Classes
1 Propellants and Fuels
101: Cryogens, 102: Hypergols, 103: Nuclear Fuel, 104: Petroleum Fuels,
105: Other Fuels, 106: Green Propellants
2 Crew Provisions
201: Water and Support Equipment, 202: Food and Support Equipment, 203: Gases,
204: Hygiene Items, 205: Clothing, 206: Personal Items
3 Crew Operations
301: Office Equipment and Supplies, 302: EVA Equipment and Consumables,
303: Health Equipment and Consumables, 304: Safety Equipment,
305: Communications Equipment, 306: Computers and Support Equipment
4 Maintenance and Upkeep
401: Spares and Repair Parts, 4011: Pressurized Spares, 4012: Unpressurized Spares,
4013: Repair Parts, 402: Maintenance Tools, 403: Lubricants and Bulk Chemicals,
404: Batteries, 405: Cleaning Equipment and Consumables
7 Waste and Disposal
701: Waste, 702: Waste Management Equipment, 703: Failed Parts
In addition to a class of supply, resources are described with a unit of measurement, unit mass (kilograms
per unit), and unit volume (cubic centimeters per unit). These additional properties allow customizable
resources with arbitrary units.
There are three subclasses of resources used in analysis: generic, continuous, and discrete. All three share
the same interface enabling a mixture of different levels of abstraction in a single analysis. For example, it
is possible to use generic resources for crew demands, continuous resources for propellant demands, and
discrete resources for spares demands.
Generic Resources
Generic resources are the simplest type of resource which is directly tied to a specific class of supply.
They are most useful for performing high-level analysis using approximations for demands by class of
30
supply. Generic resources are always expressed in units of kilograms with a unit volume estimated from
the class of supply‟s average density, though volume is can be disregarded for high-level analysis.
Continuous Resources
Continuous resources are customized resources having continuously-variable quanta such as propellants,
liquids, and gases. Continuous resources are usually measured in units of kilograms but other units can be
defined. Zero-mass resources such as electricity could be modeled using continuous resources having zero
unit mass and volume. Table 7 lists examples of continuous resources.
Table 7: Continuous resource examples.
Continuous Resource COS Units Unit Mass
[kg/unit]
Unit Volume
[cm3/unit]
Ammonium Perchlorate Fuel 105 kg 1 500
Potable Water 201 kg 1 1000
Lunar Samples 603 lb 0.45 200
Electricity 105 kW-h 0 0
Discrete Resources
Discrete resources are customized resources having indivisible units. Examples of discrete resources may
include supply items such as a packaged meal or a toothbrush, or components used as spare parts.
Depending on the level of analysis, the line between discrete resources and elements may be blurred as
both represent indivisible objects. Resources, however, are the product of demand rather than
instantiation, and only elements can generate demands.
The units for discrete resources are always “item” though the unit mass and unit volume can be set to
correspond to the individual items. Utilization of discrete resources is often different from continuous
resources, as the item may not actually be used in the sense that it no longer exists (e.g. maintenance
tools). In this sense, utilization is understood to be the demand for the existence of a new item. Table 8
lists examples of discrete resources.
Table 8: Discrete resource examples.
Discrete Resource COS Units Unit Mass
[kg/unit]
Unit Volume
[cm3/unit]
Leisure Book 206 item 0.15 700
Repair Wrench 402 item 0.24 200
Computer Motherboard 4011 item 0.65 4500
31
2.3 Element Model
Elements have a rich hierarchy to represent a variety of specialty capabilities. Elements are each uniquely
identified upon instantiation in a campaign. They may contain parts, incorporating resources to help drive
demands for spares and other activities such as repair or scavenging, and may have one or more
operational states of several categories, enabling multi-modal demand generation.
Similar to resources, elements are also assigned a class of supply. Table 9 lists classes of supply
commonly assigned to elements. Elements also have a specified mass, volume, and stowage environment
(pressurized or unpressurized) which determines allowable nesting relationships.
Table 9: Common element classes of supply.
COS Description and Sub-Classes
5 Stowage and Restraint
501: Cargo Containers and Restraints, 502: Inventory Management Equipment
6 Exploration and Research
601: Science Payloads and Instruments, 602: Field Equipment, 603: Samples
8 Habitation and Infrastructure
801: Habitation Facilities, 802: Surface Mobility Systems,
803: Power Systems, 804: Robotic Systems,
805: Resources Utilization Systems, 806: Orbiting Service Systems
9 Transportation and Carriers
901: Carriers, Non-propulsive Elements, 902: Propulsive Elements
2.3.1 Element Hierarchy
Figure 9 illustrates a hierarchy of elements developed to account for specialized functions while allowing
as many common attributes as possible. The base element generates demands for resources using parts
and states. Subclasses of elements include human agents (crew), robotic agents (reserved for future
expansion), resource containers and tanks containing quantities of resources, and carriers containing other
elements. Subclasses of carriers include propulsive vehicles which can traverse space edges using
propulsive burns and surface vehicles which can traverse surface edges at a specified speed.
32
Figure 9: Element hierarchy. A hierarchy of elements provides specialized capabilities
extensible for future expansion while establishing a common set of inherited properties.
The hierarchy of elements is likely to be expanded with future development by building on the existing
functionality. For example, if low-thrust (ion propulsion) vehicles were desired, a new sub-class of the
Carrier interface could be created similar to the existing Propulsive Vehicle.
Elements
The simple element forms the base of the element hierarchy. Elements cannot contain other elements or
resources or traverse edges under their own power but may generate or contribute to demands for
resources. Common objects modeled as elements include infrastructure objects such as solar panels, radio
communication links, and scientific payloads.
Human Agents
Human agents represent crew members. Presently, the only difference from basic elements is that human
agents specify a fraction of time that the agent is available (typically estimated at 2/3). Human agents
count towards crew capacity constraints for carriers, and are also accounted for in several measures of
effectiveness (see Section 3.4.2).
Future development is targeted at the “agent” interface, which both human agents and future robotic
agents will implement. Human and robotic agents are distinguished from other elements in that they will
be able to perform tasks, such as preventative or corrective maintenance.
Resource Containers
Resource containers are elements with the primary purpose of containing one or more types of resources
up to a mass or volume limit. Table 10 lists resource containers examples including cargo transfer bags
(CTBs) and SPACEHAB Oceaneering Space Systems (SHOSS) boxes for spare parts.
33
Table 10: Resource container examples.
Resource
Container COS
Tare Mass
[kg]
Volume
[m3]
Max Cargo
Mass [kg]
Max Cargo
Volume [cm3]
CTB 501 1.8 0.0529 45.4 52900
Double CTB 501 3.6 0.1058 90.7 105800
Half CTB 501 1.0 0.0248 4400 24800
SHOSS Box 501 120 0.4444 200 444400
Resource Tanks
Resource tanks are elements similar to resource containers but can only contain one type of resource up to
a set amount of units of resource. Since the capacity does not depend on mass or volume limits, resource
tanks can be used for zero-mass resources. Typical applications for resource tanks are for fuel, propellant,
water, or gas storage. Table 11 provides examples of a few resource tanks.
Table 11: Resource tank examples.
Resource Tank COS Tare Mass
[kg]
Volume
[m3]
Resource Max Amount
[units]
Gas Tank 501 108 2.75 Generic
COS 203
100 [kg]
Gas Tank
Derivative
501 10.8 0.275 Generic
COS 203
10 [kg]
Liquid Tank 501 34.4 0.0748 Generic
COS 201
74.8 [kg]
Liquid Tank
Derivative
501 11.5 0.0249 Generic
COS 201
24.9 [kg]
Notional Battery 501 90 0.05 Electricity 15.6 [kW-h]
Carriers
Carriers are elements containing contain, or carry, other elements up to a capacity limits for crew size,
mass, and volume. Though similar in function to a vehicle or habitat, the term “carrier” was selected
because “vehicle” inferred the capability of independent movement and “habitat” inferred capacity for at
least one crew member. Carriers provide a cargo environment (pressurized or unpressurized) which
defines what elements can be carried. As each carrier may only have either pressurized or unpressurized
cargo, multiple carriers can be used to model multiple cargo environments. Common applications for
carriers include habitation modules, non-propulsive in-space vehicles, and logistics and stowage modules.
Carriers are also often used for abstracted elements used in flight transports.
34
Surface Vehicles
Surface vehicles are a subclass of carriers having surface mobility capabilities to traverse surface edges.
They also have an integrated resource tank to carry fuel independent from cargo capacities.
Propulsive Vehicles
Propulsive vehicles are a subclass of carriers having propulsion capabilities needed to traverse space
edges. Propulsive vehicles have up to two integrated resource tanks, one to supply an Orbital
Maneuvering System (OMS) engine, and one to supply a Reaction Control System (RCS) engine.
Propulsive vehicles may have one or both systems, availability specified by a specific impulse, and may
share a common fuel tank for both systems.
2.3.2 Element Parts
Parts represent quantities of resources applied in an element. Though assigning parts to an element does
not change spares demands alone, the parts list may be used to drive demands in sparing models.
Additionally, parts can be assigned a mean time to failure (MTTF) to further inform detailed sparing
models. If a part is designated as repairable, it is assigned a mean repair time (MRT) and mean repair
mass (MRM) to cover low-level components and tooling, though the decision of whether to repair is
made at the campaign level.
Parts are inferred as common if multiple elements are assigned the same underlying resource. Common
parts are advantageous for spares pooling and scavenging from decommissioned elements. Supplying an
exhaustive list of parts is often not feasible, especially for abstracted elements, high-level studies, or yet-
to-be-designed components. In practice, a few notional parts
that exhibit desired features such as commonality and
repairability can be used to perform analysis. Figure 10
illustrates a notional case where two elements, a power plant
and a lander, share a common power supply resource type. In
this case, 10% of the power plant by mass is common with 20%
of the lander by mass.
2.3.3 Element States
States model the operational capability of an element, allowing
changes to the demand profile for an element during
exploration. Element operational states fall under one of the
Figure 10: Part commonality with shared
resource types. The Power Plant and
Lander elements share a common Power
Supply resource.
35
five following categories:
Active: nominal operational level
Special: short-duration increased operational level, often used to represent EVAs
Quiescent: reduced operational level
Dormant: minimal operational level, often used for elements before activation or deployment
Decommissioned: permanent non-operational level; the element‟s parts are available to scavenge
An element can have any number of possible states and one current state which changes through
undergoing a reconfiguration. Each state specifies a set of demand models that generate demands for the
element while it is in that state. Figure 11 depicts a typical reconfiguration sequence.
Figure 11: Element reconfiguration sequence. An element may experience several operational
states throughout a campaign including dormant before delivery, active or quiescent during
operation, and decommissioned state to signal availability of parts for scavenging.
2.3.4 Element Demand Models
Element demand models generate demands for resources during simulation from the perspective of each
element, which are then aggregated by node. This allows multi-fidelity element-level demand models to
supplement the traditional method of modeling demands using parameterized functions on a global
perspective (see Section 3.2.1). As element states are changed over the course of a campaign, different
demand models are utilized to create a complex composite demand model that would be difficult to
achieve from a global perspective.
Two examples of simple element demand models are presented below. Additional demand models are
likely to evolve in future development to represent more complicated phenomena such as boil-off, shelf
life, and advanced sparing models.
Linear Demand Model
The linear demand model produces demands for a set of resources linearly with time. Its cumulative
demands for resource i can be expressed as a function of the simulation time t in Eq. (1).
36
tbatD iii )( (1)
However, as the simulation runs in discrete time intervals, the
demands generated at each event execution are expressed as a
function of the number of days since the previous event
execution Δt, shown in Eq. (2). Functionally, the check for the
first aggregation is performed using a Boolean flag that is
initialized to zero and set to one after the first aggregation.
otherwise
naggregatiofirst if )(
tb
tbatd
i
ii
i (2)
The resulting demands resemble a step function, illustrated in Figure 12, though the demands are being
produced at a linear rate. In practice, the constant terms ai are used for fixed demands such as tools,
safety, or health equipment and the linear terms bi are used for recurring demands such as food, water or
oxygen for human agents, or electricity for other elements.
Sparing by Mass Demand Model
The sparing by mass demand model is a simple sparing model assuming the mass of spare parts
demanded by an element annually is proportional to a percentage of its dry mass. The spare parts rates
can be split by environment into pressurized spares (COS 4011) and unpressurized spares (COS 4012),
lumped into an unspecified category (COS 401), or a combination of both.
The basic formulas for deriving spares demands is an extension of the linear demand model generating
demands for generic resources. Given rates ri representing unspecified, pressurized, and unpressurized
rates for class of supply i = 401, 4011, and 4012 and the element mass melement, the generic resource
demands di,generic aggregated after Δt days of simulation time are shown in Eq. (3).
25.365
)(,
tmrtd elementiigeneric (3)
If an element has a parts list defined, the sparing by mass model can optionally use non-generic resource
types in proportion to each part‟s mass fraction of the overall element, separating unspecified, pressurized
Figure 12: Discrete event time effect
on linear demands. Discrete time
simulation aggregates linear demands at
each event execution.
37
and unpressurized demands. Figure 13 illustrates a parts list and corresponding resource type breakdown
for a notional 100 kilogram surface rover with three specified parts totaling 50 kilograms.
Figure 13: Sparing by mass using a parts list. The sparing by mass demand model optionally can access an
element’s part list to generate unspecified, pressurized, and unpressurized spares demands.
There are two fractions that drive the spares distribution when using parts lists. First, the element class of
supply fraction fcos,i identifies the portion of an element comprised of class of supply i. For all elements,
fcos,401 = 1, and both fcos,4011 and fcos,4012 are calculated by finding the fraction of element mass not
accounted for by parts of other classes of supply. In the example in Figure 13, fcos,4011 = 0.6 and fcos,4012 =
0.75. Second, the part fraction fpart,j identifies the portion of an element comprised of part j. In the
example in Figure 13, fpart,1 = 0.25, and fpart,2 = 0.10 and fpart,3 = 0.15 for the drive train, air filter, and
electronics respectively.
Using the two fractions, Eq. (4) shows the modified functions to calculate demands for each part type j.
All part types are considered instances of class of supply 401, but only pressurized parts are instances of
class of supply 4011, and only unpressurized parts are instances of class of supply 4012.
otherwise 0
supply of class of instancean is part if 1 where
25.365)(
,
,,cos,,
ijX
tXffmrtd
ji
i
jijpartielementijpart
(4)
Finally, Eq. (5) shows the adjusted generic resource demands after accounting for parts demands. If no
parts list is defined, Eq. (5) reduces to Eq. (3).
38
25.365
1)( ,cos,,
tffmrtd
j
jpartielementiigeneric (5)
Table 12 lists the breakdown of annual demands for the notional surface rover example by class of supply
and part under spares demands rates of 5% unspecified, 10% pressurized, and 5% unpressurized. As most
parts will represent discrete resources, fractional units may be aggregated using an item discretization
policy and aggregation settings, handled during simulation (see Section 3.3.2).
Table 12: Annual demands using a sparing by mass model with a parts list.
Category \ Resource [kg]
Generic
COS
401
Generic
COS
4011
Generic
COS
4012
Electronics Air
Filter
Drive
Train Total
Unspecified Demands
(5% of element mass / year) 2.50 0 0 0.75 0.50 1.25 5.00
Pressurized Demands
(10% of element mass / year) 0 8.33 0 0 1.67 0 10.00
Unpressurized Demands
(5% of element mass / year) 0 0 3.33 0 0 1.67 5.00
Total 2.50 8.33 3.33 0.75 2.17 2.92 20.00
39
3 Campaign Modeling and Analysis
A space exploration model is comprised of three components: a network, missions, and a cargo manifest.
The network describes the nodes and edges that are accessible during the exploration. The mission
sequence defines the sequence of events to create and move elements within the network. The cargo
manifest accounts for demands occurring during the mission sequence by manifesting resource containers
into carriers. Figure 14 illustrates the process used to model and analyze a campaign, including iteration
loops in the manifest definition to allow logistics strategies such as pre-positioning.
Figure 14: Campaign modeling and analysis. Campaigns are modeled in three parts: a network,
missions, and a manifest. When the campaign is both spatially feasible and logistically feasible,
analysis can provide visualizations and measures of effectiveness.
3.1 Network Model
The network model uses a set of nodes and edges to define the space over which the campaign will
operate. Any combination of nodes and edges from any
available subclasses are allowed provided every edge
has a defined origin and destination node. The network
is referred to as a time-expanded network, as all
locations are expressed in time-invariant forms. A time-
expanded network is visualized in a plot of time versus
location, shown in Figure 15. This plot is referred to as
a bat chart when showing elements moving through the
network due to similarities in appearance to bats
perching on a ceiling.
Figure 15: Network bat chart. A bat chart
shows a time-expanded network and element
movement over time.
40
3.2 Mission Model
Missions make up the majority of a particular space exploration campaign definition. Each mission
contains a set of optional demand models to generate aggregated demands and a list of events to drive
simulation. Missions can be checked for spatial correctness using a spatial simulator.
3.2.1 Mission Demand Models
Mission demand models generate aggregated demands from a global perspective, as opposed to element
demand models which generate demands for individual elements (See Section 2.3.4). They are the
traditional choice for campaign analysis and are typically parameterized by a large number of input
variables. The consumables model used in SpaceNet 1.4 used parameters similar to those in Table 13 as
inputs to equations for demand estimation. Many of the mission and environment parameters can be
inferred by inspecting existing elements and planned mission events in a mission model, however some
including the ECLSS parameters are outside the existing model scope and would be required as inputs.
Table 13: Crew consumables mission demand model parameters.
Mission Parameters Environment Parameters ECLSS Parameters
Crew Size
Surface Duration
Non-crewed Surface Duration
Transit Duration
Supply Reserves Duration
Number In-transit EVAs
Number Surface EVAs
Crew per EVA
Average EVA Duration
Habitat Volume
Habitat Air Pressure
Habitat Leak Rate
Airlock Volume
Airlock Efficiency
Waste Water Recovery Rate
Solid Water Recovery Rate
Brine Recycling (y/n)
Brine Recycling Rate
Sabatier Reaction (y/n)
Electrolysis (y/n)
Methane Reformer (y/n)
Recover EVA CO2 (y/n)
Launder Clothes (y/n)
ISRU O2 Production Rate
In practice, mission demand models work well for aggregating mission impacts on demands but struggle
to account for element-related impacts relying on parameters describing each element type in detail. Most
campaign analysis will use a combination of both mission and element demand models to represent
aggregated and individual demands.
3.2.2 Mission Events
Each mission defines a list of events to control the instantiation and movement of elements to accomplish
exploration operations. There are a wide variety of events to represent the various activities that occur
during a space exploration campaign, however there are a set of seven core events from which all events
41
can be derived. Four events, shown in Table 14, relate to the lifecycles of elements and three events,
shown in Table 15, relate to the usage of resources. Composite events, shown in Table 16, automate the
use of one or more core events to perform a complex event with minimal input.
Table 14: Element-based core events list.
Event Name Description Spatial Error Conditions
Instantiate
Elements
Instantiates elements at a
location or nested inside a
carrier.
The target carrier has not been instantiated or does not exist
The target carrier does not have sufficient capacity
An element to be instantiated already exists in the simulation
Move Elements Instantaneously moves
elements to a new location
or carrier.
The target carrier has not been instantiated or does not exist
The target carrier does not have sufficient capacity
An element to be moved has not been instantiated or does not
exist at the expected location of origin
Remove Elements Permanently removes
elements from the scope of
simulation.
An element to be removed has not been instantiated or does
not exist at the expected location
Reconfigure
Elements
Changes elements‟
operational state.
An element to be reconfigured has not been instantiated or
does not exist at the expected location
An element to be reconfigured does not contain the target
reconfiguration state
An element to be reconfigured is already in an unchangeable
decommissioned state
Table 15: Resource-based core events list.
Event Name Description Spatial Error Conditions
Add Resources Adds resources to an
existing resource container.
The resource container has not been instantiated or does not
exist at the expected location
The resource container has insufficient capacity
Transfer Resources Transfers resource from an
existing origin resource
container to a co-located
destination resource
container.
The origin or destination resource container has not been
instantiated or does not exist at the expected location
The origin resource container has insufficient resources
The destination resource container has insufficient capacity
Remove Resources Creates a demand for
resources originating from a
target element.
The target element does not exist or does not exist at the
expected location
The resource container has insufficient resources
Each mission event is assigned an execution time relative to the start of the mission and a priority
between one and five. The priority is used to assign preference to simultaneous events where dependence
is required. For example, elements may be instantiated at a node at the same time a transport begins, but
clearly the instantiation must occur first. The seven core events and propulsive burns are modeled as
instantaneous events at a single location. EVAs and explorations are modeled as processes occurring over
42
a set duration at a single location. Space, surface, and flight transports are modeled as transports occurring
over a set duration between an origin and destination location.
Table 16: Composite events list.
Event Name Description Events Used
Propulsive Burn Performs an impulsive burn using one or more propulsive vehicles‟
OMS or RCS engines with staging to achieve a target delta-v.
Propellant is expended according to the rocket equation.
Remove Resources
Remove Elements
Space Transport Transports a set of elements (the stack) across a space edge using a
series of propulsive burns to achieve required delta-v specified by the
space edge.
Move Elements
Propulsive Burn
Surface Transport Transports a surface vehicle and any nested elements across a surface
edge at a constant speed and given duty cycle. The surface vehicle is
reconfigured to an optional transport state before the start of the
transport.
Reconfigure
Elements
Move Elements
Flight Transport Transports a set of elements (the stack) up to a nested mass and crew
capacity across a flight edge.
Move Elements
Extravehicular
Activity (EVA)
Schedules an extravehicular activity during which crew members are
reconfigured to an EVA state if specified and moved external to the
habitat (carrier). After the EVA duration, crew members are returned
to the habitat and reconfigured to the previous state.
Reconfigure
Elements
Move Elements
Exploration Schedules a specified number of EVA events having equal duration
and the same crew over an exploration period. EVAs are scheduled
with equal time separation before and after all events.
EVA
Both OMS and RCS propulsive burns utilize a specified burn/stage sequence and the rocket equation to
calculate demands for fuel using the assumption of impulsive burns. Eq. (6) is used to calculate the
maximum delta-v achievable Δvachievable for a stack of mass mstack and vehicle with fuel mass mfuel and
specific impulse Isp. If the achievable delta-v is less than the target delta-v, all fuel is consumed, any
specified elements are staged, and the next propulsive vehicle in the sequence repeats the same analysis.
fuelstack
stack
2achievable logs
m81.9
mm
mIv sp (6)
Alternatively, if there is excess fuel in a particular propulsive vehicle, the remaining delta-v is decreased
to zero and Eq. (7) determines the amount of fuel to consume in the burn mburn.
2
target
stackburnsm81.9
exp1spI
vmm (7)
43
3.2.3 Spatial Simulator
It is important to identify error conditions as soon as possible as missions are populated with events. The
types of errors occurring during this phase of campaign modeling are called spatial errors, meaning the
error is related to the location or placement of one or more elements in the system state. The most
common spatial errors occur when elements do not exist at an expected location or a carrier element has
insufficient capacity to nest a required element. Since spatial errors are only perceived at event execution
time and the system (network) state before each event depends on all previous events, the entire campaign
must be simulated to determine the presence of any spatial errors. A spatial simulator quickly executes the
events without demand satisfaction for this purpose.
The spatial simulator uses the information contained within the campaign to build an initial network state
and an event stack sorted by execution time. Each event is sequentially executed, resulting in an updated
network state and new events added to the stack for composite event execution. Any spatial errors
identified during execution are logged for debugging. Figure 16 illustrates the general process of
simulating a campaign using a spatial simulator.
Figure 16: Spatial simulation process. The spatial simulator sequentially executes events,
updating the network state and adding events when necessary, and may generate spatial errors.
3.3 Manifest Model
The manifest model defines a set of resource containers packed with resources and a list of cargo transfers
between transport carriers to ensure resources are in place to satisfy demands during the exploration
campaign. Figure 17 outlines the process used to generate a manifest, including iteration loops required to
share containers and create multi-transport manifests.
44
Figure 17: Manifest model creation sequence. Manifesting requires the generation of demands,
aggregated to the location and time of demand. Resources are then packed into containers which
subsequently are manifested onto carriers for transport.
First, demands must be generated by executing the campaign with a demand simulator which logs
unsatisfied demands for resources. Second, raw demands are aggregated to simplify the delivery of
resources to locations at particular times. Third, the aggregated resources are divided and packed into
containers. Finally, the containers are manifested onto a sequence of carriers to bring the resources to the
time and location at which it is demanded. Iteration loops exist if containers hold resources used at
different times or places or if containers must be manifested onto more than one transport.
The manifesting model presented in this section only provides a representation of the manifest model
within a space exploration campaign, not necessarily a method to create a manifest to meet a particular
objective. The implementation of the manifesting model in SpaceNet 2.5, detailed in Section 4.3.5,
provides a heuristic algorithm which attempts to find a valid, but sub-optimal, manifest. Appendix B
presents a body of parallel research into the optimal manifesting problem, which uses a similar modeling
framework to find an optimal manifest subject to a policy, strategy, or other objective function.
3.3.1 Demand Simulator
The demand simulator is a more detailed version of the spatial simulator discussed in Section 0. The main
addition is a demand satisfaction sequence resulting in a list of unsatisfied demands, as shown in Figure
18. Most demands are yet unsatisfied and logged before creating the cargo manifest.
45
Figure 18: Demand simulation process. The demand simulator executes events, updating the
network state and adding events as necessary, may generate spatial errors, and logs unsatisfied
demands for analysis and manifesting.
A demand satisfaction sequence generates and attempts to satisfy demands during execution of the
campaign events, adjusting for discretization, repair, and scavenging. The demand satisfaction sequence
includes the following six steps:
1. Generate raw demands from demand models
2. Discretize demands per discretization and item aggregation policies
3. Repair demands per repair policy
4. Scavenge resources if available per scavenging policy
5. Satisfy remaining demands using co-located resources
6. Report unsatisfied demands with (optional) packing overhead mass
Demand Generation
Demands for resources originate from mission and element demand models. Since the two types of
demand models operate on different levels, demand generation differs slightly. Mission demand models
are queried before each mission starts for a list of resources to support the mission. Element demand
models selected by the current element operational state are queried before each event execution for a list
of resources demanded since the previous event execution. In some cases, there may be no elapsed time
from the previous event, in which case there may be no additional demands. Co-located demands for the
same resource are combined so there may is one demand for each resource from a single location.
Discretization and Item Aggregation
The time between subsequent events may often be shorter than the time required for generation of a unit
for discrete resources (items). In an effort to manage the discretization process, two policies guide the
46
generation of whole items. The discretization policy determines how continuous demands are grouped
into units and the item aggregation policy determines when the discrete units are resolved.
Figure 19 illustrates three options for the discretization policy: no discretization, by element, and by
location. No discretization treats discrete resources as continuous resources and no further considerations
are taken. Discretization by element aggregates discrete items separately for each element, independent of
location. Discretization by location aggregates items separately for each location, allowing for spares
pooling between co-located elements with common demands. The choice of discretization policy can help
smooth demands for common spare parts at a single location.
Figure 19: Impact of discretization policy. Sample demands for two elements with common
discrete demands. AP is the aggregation policy value. (a) No discretization treats the demands as
continuous. (b) Discretization by element aggregates items on a per-element basis. (c)
Discretization by location aggregates items on a per-location basis, enabling spares pooling to
locally decrease and smooth demands.
The aggregation policy, ranging from 0 to 1, sets the point at which continuous quantities are aggregated
into items. The policy value determines the amount of a discrete resource to be generated before a whole
unit demand is aggregated. Figure 20 illustrates several aggregation policy values. A policy level of 0.0
aggregates items at the first partial demand (equivalent to the ceiling function) while a policy level of 1.0
aggregates items after unitary demands are accumulated (equivalent to the floor function). Most analyses
use an aggregation policy of 0.5 to balance the over-estimating and under-estimating behaviors.
47
Figure 20: Impact of aggregation policy. The item aggregation policy determines at which point
demands are aggregated into whole units. (a) No discretization treats demands as continuous. (b)
A policy of 0.0 aggregates items at the first demand. (c) A policy of 0.5 aggregates items when half
units are accumulated. (d) A policy of 1.0 aggregates items after unit demands.
Repair
Repair activities provide a trade between crew repair time and spare parts mass. Repair activities are
dictated by a repair list of items targeted for repair during each crewed mission (for implementation
details, see Section 4.3.4). During the repair cycle, the demand for a repaired item is replaced by any
derived resources required to perform the repair.
Scavenging
If enabled, resources can be scavenged from co-located elements previously having entered a
decommissioned state (see Section 2.3.3). The scavenging process reduces the amount of resources
demanded by the amount of available parts contained in co-located decommissioned elements until there
are no remaining resources to scavenge.
Demand Satisfaction
Any remaining demands after discretization, repair, and scavenging attempt to be satisfied by existing
resources contained in elements co-located at the node or edge at which the demand was generated. If a
demand originates from an element, it will be recursively inspected for any available resources including
nested elements before all other co-located elements are recursively inspected for available resources. All
unsatisfied demands are logged and supplied to the demand aggregation phase.
48
3.3.2 Demand Aggregation
Demands are aggregated into groups for further
processing after generation by the demand
simulator. The goal of demand aggregation is to
group demands having similar spatial and temporal
characteristics as to how they may be satisfied via
manifesting. This is accomplished with the
introduction of exploration periods, defined as the
time serviced by a particular transport, i.e. the
interval between subsequent arrivals of transports or
the end of a campaign at a specific node. Figure 21
demonstrates the exploration period definition for a sample campaign of eight transports.
Every demand in a campaign can be aggregated to either a transport (demands at edges) or an exploration
period (demands at nodes). Although all demands within a transport must be self-supplied, demands
within an exploration period can be supplied by any transport arriving at the same node at or before the
start of the period. By definition, there are no transport arrivals in the middle of an exploration period,
providing a common requirement for all inter-period demands and greatly simplifying the problem. Using
these rules, resource containers can be created and packed with resources and manifested onto transports.
3.3.3 Demand Packing
The process of demand packing assigns quantities of resources to individual resource containers. Figure
22 shows the two-step process to packing. First,
aggregated demands are divided into groups
corresponding to packing separation. Second,
resource containers are packed with cargo from one
or more packed demand groups according to mass,
volume, and environmental constraints and delivery
constraints derived from other resources sharing the
container. The result of the demand packing process
is the dual map from aggregated demands to
demands as packed to resource containers.
Figure 22: Demand packing. Aggregated demands
are split into groups corresponding to how the
demands are packed within individual resource
containers.
Figure 21: Campaign transports and exploration
periods. Each transportation arrival marks the start
of its corresponding exploration period.
49
3.3.4 Container Manifesting
The process of manifesting assigns resource
containers to carriers. Manifests include a sequence
of transports completing a reverse transversal from
the demand location to a source node. As containers
are manifested on carriers, they are subsequently
demanded at the exploration period preceding the
transport, requiring additional manifest steps to reach
a source node. Figure 23 shows example multi-
transport manifesting options for a demand from
exploration period three.
3.4 Campaign Analysis
After the manifest has been created, the campaign is ready for the analysis phase. At this point, all
demands should be satisfied by the manifested resource containers. The campaign simulator creates
visualizations and computes various measures of effectiveness.
3.4.1 Campaign Simulator
The campaign simulator is an extension of the demand simulator. The completed cargo manifest is
processed into scheduled events to create, fill, and transfer resource containers. Additional data logs are
maintained to record the system state after each event execution and tabulate activities such as scavenging
and repair. Finally, a set of measures of effectiveness are maintained to evaluate the entire campaign.
Figure 24: Campaign simulation process. The campaign simulator executes events, updating the
network (system) state and adding events as necessary, and logs measures of effectiveness.
Figure 23: Multi-transport manifesting.
Containers manifested on transport 3 (T3) are
demanded during exploration period 2 (E2) and
must be supplied by either T1 or T2.
50
Any spatial errors or unsatisfied demands occurring during the simulation are noted for corrective action,
though the simulation will still proceed. This most commonly results from a manifesting action affecting
a mission event, e.g. the addition of extra cargo mass causes a propulsive infeasibility.
3.4.2 Measures of Effectiveness
The NASA Systems Engineering Handbook identifies measures of effectiveness (MOEs) as “… the
„operational‟ measures of success that are closely related to the achievement of mission or operational
objectives in the intended operational environment” [28]. A summary of the MOEs used in this campaign
analysis is provided in Table 17. Detailed MOE formulations are described in previous publications [10].
Many MOEs are carried over from past work, though the calculation procedure is changed from post-
processing to live-logging. In a live-logging environment, a log entry is created for contributions to each
of the MOEs before each event is executed in the campaign simulator. For example, if a transport event
moving elements from Kennedy Space Center into low Earth orbit is to be executed, its contributions to
the total launch mass and up-mass capacity utilization are logged. The resulting logs can be used to
recreate the evolution of the MOEs over the course of a campaign.
Table 17: Campaign measures of effectiveness.
Measure of Effectiveness Units Description
Crew Surface Days crew-days The total number of crew-days over all non-Earth
surface nodes.
Crew Corrective Maintenance Time crew-hours The total number of crew-hours spent on corrective
maintenance (repair) activities.
Exploration Mass Delivered kg
The total mass of exploration items (COS 6) and
surface infrastructure (COS 8) delivered to all non-
Earth surface nodes.
Total Launch Mass kg The total mass transported from Earth‟s surface.
Up-mass Capacity Utilization - Fraction of available mass capacity utilized for
cargo on transports from Earth‟s surface.
Down-mass Capacity Utilization - Fraction of available mass capacity utilized for
cargo on transports to Earth‟s surface.
Exploration Capability kg-crew-days
The dot product of crew surface days and
exploration mass, resulting in a measure of the total
capability for crew to perform exploration.
Relative Exploration Capability - The amount of productive exploration per kilogram
of mass launched as compared to that of Apollo 17.
51
4 SpaceNet 2.5 Implementation
SpaceNet 2.5 is a Java program implementing the models discussed in
Chapters 2 and 3. It was developed using user-centric design philosophies to
improve its usability and efficiency. The SpaceNet graphical user interface
(GUI) allows the user to build, edit, and analyze exploration campaigns
without detailed knowledge of the underlying models. SpaceNet also
provides visualizations and feedback to simplify the campaign creation
process and quickly identify and reduce the number of simulation errors.
4.1 Development and Release
Development on SpaceNet 2.5 started in July 2008 using user-centric methodologies to iterate on
concepts and design prototypes. MIT hosted a SpaceNet workshop on December 4-5 2008 which
established core concepts including the element hierarchy, element operational states, and element
demand models. Through January 2009, development focused on maturing concepts with prototypical
user interfaces and sample simulation implementations.
From January through June 2009, the focus of development turned to design and implementation of the
graphical user interface and maturing the campaign simulation to allow manifesting of demands. A
review was hosted at JPL on March 24 2009 at which time a prototype was demonstrated including
demand simulation. A NASA-specific version of SpaceNet 2.5 was delivered on June 30 2009 along with
documentation including a User‟s Guide and Quick Start Tutorial. Between July and September 2009,
non-publicly available components were removed from the public branch of SpaceNet, resulting in the
public release on October 1, 2009.
Development after October 2009 focused on usability improvements and expanded support for data
management including a data editor. SpaceNet 2.5 R2 is scheduled for release in late 2010.
Figure 26: SpaceNet 2.5 development timeline. SpaceNet 2.5 R1 was released after initial
delivery to NASA JPL. SpaceNet 2.5 R2 is currently under development.
Figure 25: SpaceNet logo.
52
4.1.1 Development Methodology
Improve its usability and efficiency was one of the early development goals for SpaceNet. The high-level
goal of usability, however, is coupled with the low-level goal of modeling, so several prototype iterations
of increasing fidelity were created and presented to representative users during the development cycle.
The first prototypes included sketches and screen mockups to display the process of building and
analyzing a campaign. Figure 27 shows an example mockup from an early design cycle. Later prototypes
used spreadsheets and forms to provide feedback and visualizations during the campaign design process.
Figure 27: Early SpaceNet GUI prototype mockup. Prototype mockups assist development
iterations by walking through the campaign definition without time-intensive coding investment.
4.1.2 Development Tools
A SpaceNet server running LAMP (Linux, Apache, MySQL, PHP) provided development tools to enable
collaboration between team members. First, the server hosted a wiki using the open-source DokuWiki
software [29], providing a central point of communication between all members of the development team.
In addition to meeting notes and documentation, the SpaceNet wiki also provided a bug reporting and
change request page summarizing and sorting feedback by level of priority (bug, high, medium, or low).
The wiki logged over 100 requests throughout development of SpaceNet 2.5 R1.
Establishing a subversion repository on the server was another crucial component for development.
Subversion (SVN) provides version control over a central code base [30]. In addition to the ability to roll
back to previous versions, developers receive updates from others in a collaborative fashion. The
subversion system logged over 1000 modifications by the release of SpaceNet 2.5 R1.
53
Finally, developers used the Eclipse Integrated Development Environment (IDE) for writing and
debugging code [31]. Eclipse provides continuous-compilation of a project to identify errors and
refactoring support for automating changes. Extensions to the Eclipse platform also provide direct
integration with the SVN server.
4.1.3 Public Release
SpaceNet 2.5 R1 was released under an open source GNU General Public License (GPL) version 3 on
October 1 2009 [32]. This type of license is referred to as a copyleft license, as its terms protect the
content of the copyright holder by ensuring free access to the source code on the condition that any
derivative works must also be distributed under a similar license. Under this philosophy, a steady-state
market price of zero is achieved for the source code, though services and support may be provided for the
original or derivative works for a fee.
A project website was created on the SpaceNet server to organize information about the project and
provide a download point for potential users.* An online community was also established to provide a
point of contact between users and developers.
4.1.4 Documentation
Documentation included with distributions of SpaceNet 2.5 R1 includes the SpaceNet User’s Guide,
SpaceNet Quick Start Tutorial, and Javadoc comments. Much of the technical details from the
documentation has not been repeated in this thesis due to its dynamic nature with additional development
and provided accessibility via the SpaceNet project website.
The SpaceNet User’s Guide is targeted as a reference document rather than a tutorial. It provides an
overview of the project motivation and implementation as well as information on every component of the
SpaceNet GUI. Two appendices aid the understanding of the underlying model. The first defines the
abstract object classes used in the modeling framework. The second details the data required by the data
source to populate object definitions.
The SpaceNet Quick Start Tutorial walks through two step-by-step tutorials highlighting the analysis
techniques and campaign-building methods. The first tutorial creates a lunar sortie mission to the Lunar
South Pole focused on introducing the basics of mission events and establishing propulsive feasibility.
The second tutorial uses abstracted flight transportations to create a multi-mission campaign to the Lunar
* The project website is available at http://spacenet.mit.edu
54
South Pole focused on introducing demand models and manifesting. All required data source files as well
as completed tutorial campaign files are included with the distribution.
Javadoc files provide documentation of the detailed software implementation called an application
programming interface (API) [33]. Javadoc files are automatically generated from structured comments
placed within the source files and would typically only be used by developers. The source distribution of
SpaceNet 2.5 includes full Javadoc files detailing the model and GUI implementation.
4.2 Data Sources
Campaign analysis within SpaceNet requires many objects definitions in addition to the specific
campaign and mission architectures. Redefining objects for each campaign, however, is time-consuming.
In particular, the objects that comprise the core of the modeling framework (nodes, edges, resources, and
elements) seldom change between campaigns, prompting the usage of independent data sources that may
be used between several campaigns.
SpaceNet 1.3 and 1.4 store object definitions in spreadsheets organized into several tabs, one per object
class. As some objects require references to other objects (e.g. origin and destination nodes for each
edge), identification numbers, or keys, are used to create data relationships between tables consistent with
relational database theory. Figure 28 highlights the numerous relationships used within the interconnected
SpaceNet modeling framework.
Figure 28: Data table relationships. Data is divided across several tables to represent parent-
child relationships between objects.
Development in SpaceNet 2.0 investigated using relational databases as a replacement for spreadsheets.
The same table structure was maintained, but validation rules were added to maintain data integrity.
Database software also allows commands written in generic structured query language (SQL) and
55
performs operations faster than reading and writing to files. After preliminary user testing, however, it
was discovered that spreadsheets provided an easy, familiar interface to the data. For this reason, a
generic interface to data sources was created that does not depend on the underlying implementation. To
date, data sources have been implemented for both spreadsheets and relational databases.
The movement to an abstracted data source interface also enabled a generic data editor which provides a
graphical user interface for any data source.* This provides users with an easier way to edit and view the
data source that enforces validation and formatting rules. In addition to the data editor, element sizing
models have been introduced to aid the design of new elements for use in campaigns.
4.2.1 Data Editor
Editing data directly within a raw format such as a spreadsheet can be time-consuming and error-prone.
Spreadsheet data often loses integrity due to incomplete entries as validation rules are not enforceable.
Primary key-foreign key relationships between tables create a cascade of nested objects in both
spreadsheets and relational databases. Capitalizing on the data source abstraction, a data editor was
introduced to manage data regardless of its source implementation.
Figure 29: SpaceNet data editor user interface. The data editor provides a GUI for any data
source implementation. This example edits an edge from a spreadsheet database.
The data editor is integrated with SpaceNet to provide a graphical user interface to data sources. It
provides support for viewing, editing, deleting, and creating new entries for nodes, edges, resources, and
* Collaborator Ivo Ferreira led development and implementation of the data editor and element sizing tools.
56
elements. It helps ease the process of editing data, especially for inexperienced users who may not be
familiar with the formatting or organization of underlying data sources.
4.2.2 Element Sizing Tool
Of the domain objects in the SpaceNet modeling framework, nodes,
edges, and resources seldom change between campaigns but
elements are susceptible to modification as designs mature or trades
are considered. The ability to generate new element designs using a
sizing model was introduced as a module of the data editor.
An element sizing model provides an alternative process to generate
element specifications. For example, a crew member sizing model
may take height as an input to estimate the element mass. Similarly,
a habitat sizing model may require maximum crew occupancy and
enabled technologies to generate the element mass and volume.
Current support exists for spreadsheet-based models using
designated cell addresses for inputs and outputs. Future research and development will look into other
methods for designing elements including leveraging existing databases of similar elements to infer the
design of new elements.
4.3 Graphical User Interface
The graphical user interface (GUI) is the portion of SpaceNet which users interact with to create and
analyze campaigns. Figure 31 highlights some features of the SpaceNet GUI. The campaign panel and
five tabs guide users through the campaign definition and analysis process, broken down into six steps.
1. Set high-level campaign parameters (Campaign Panel)
2. Select network nodes and edges (Network Tab)
3. Define exploration missions (Missions Tab)
a. Set high-level mission parameters
b. Define mission events
4. Analyze demands generated for resources (Demands Tab)
5. Create campaign cargo manifest (Manifest Tab)
6. Simulate and analyze campaign (Simulation Tab)
Figure 30: Element sizing model.
An element sizing model creates
element designs using customized
model input parameters.
57
Figure 31: SpaceNet 2.5 user interface. The SpaceNet GUI guides users through the process of
modeling and analyzing exploration campaigns.
4.3.1 Campaign Panel
Positioned at the top of the SpaceNet GUI, the campaign panel modifies campaign-wide options. Inputs
include a name to reference the campaign, a starting date (epoch) from which to measure simulation time,
a reference to the user or group creating the campaign, and a short description. Two additional dialog
boxes access additional options. The „campaign options‟ dialog sets global parameters such as precision
values, constraint enforcement, demand policies, and simulation options detailed in Table 18. The „data
source‟ dialog launches the data editor to select the data source and manage the loading of data.
Table 18: Campaign options list.
Option Description Values
Time Precision [0.001,0.500] days
Demand Precision [0.001,0.500] units
Mass Precision [0.001,0.500] kg
Volume Precision [0.1,100] cm3
Enforce Volume Constraints Yes/No
Item Discretization Policy None/By Element/By Location
Item Aggregation Point First/Half/Unit Demand
{0.0, 0.5, 1.0}
Enable Scavenging Yes/No
58
4.3.2 Network Tab
The network tab serves as the interface to the network model. A network visualization displays nodes and
edges from a data source. Nodes and edges are colored based on type: surface nodes are yellow, orbital
nodes red, and Lagrange nodes purple, surface edges green, space edges red, and flight edges yellow.
Figure 32: SpaceNet network visualization. Network for a lunar campaign including surface
(yellow) and orbital (red) nodes, and surface (green), space (red), and flight (yellow) edges.
Surface nodes are mapped onto a Lambert-Azimuthal projection of planetary bodies. Orbital nodes are
mapped based on altitude and inclination. Lagrange nodes are mapped based on position. The size and
relative positioning of celestial bodies and the location of orbital and Lagrange nodes are scaled using an
arctangent function to set a maximum and characteristic distance. Edges do not correspond to actual
physical paths, but rather connect nodes with graphical arcs.
Several pre-set scenario types can be used to filter the selection of locations:
ISS: Selects Earth nodes and all connected edges
Lunar: Selects Earth and Moon nodes and all connected edges
Moon-only: Selects Moon nodes and all connected edges
Martian: Selects Earth, Moon, and Mars nodes and all connected edges
Mars-only: Selects Mars nodes and all connected edges
Solar System: Selects all available nodes and edges
4.3.3 Missions Tab
The missions tab defines the sequence of missions and provides top-level visualizations for the campaign
shown in Figure 33. The process bat chart illustrates mission event sequences and the element bat chart
illustrates the movement of individual elements through the time-expanded network.
59
a)
b)
Figure 33: Campaign visualizations. (a) The process bat chart displays mission events,
processes, and transports in time-expanded network. (b) The element bat chart displays individual
element movement in time-expanded network.
Each mission supplies several inputs including a name, a starting date (epoch), and origin and destination
nodes from which to calculate net transport and exploration durations. Also, mission demand models can
be attached to calculate aggregated mission demands (see Section 3.2.1), and mission events are defined
to drive the simulation (see Section 3.2.2).
Spatial simulation is continuously executed as mission events are created to determine the system state at
the time of each new event, identify available elements and quickly alerting of any error conditions.
Figure 34 illustrates error messages displayed in the mission event list and in the event dialog.
a)
b)
Figure 34: Spatial simulation errors. Spatial simulation errors display (a) in the mission events
list and (b) in the mission event dialog to alert the user of error conditions.
4.3.4 Demands Tab
The demands tab does not correspond to a specific component of the modeling framework, but rather
provides visualizations and analysis capabilities for the demands generated during a campaign. Options
exist to set demand-related campaign parameters, including discretization, aggregation, and sparing
policies. Other options including enabling or disabling estimates for container masses and consumption of
60
existing resources only change the resulting demand visualizations. There are a series of five
visualizations and analysis tabs.
Scenario Feasibility Visualization
The scenario feasibility visualization plots the cumulative capacity of all source transports (transports
originating from a source node) and estimates of cumulative demands for resources over the entire
campaign. It is a necessary but not sufficient condition that the capacity exceeds the demand, as local
infeasibilities may still exist at specific nodes in time. In practice, it serves as a first check for identifying
infeasible campaigns, however, if the campaign contains only a single destination the scenario feasibility
visualization is a sufficient indication of feasibility. Figure 35 illustrates scenario feasibility visualization
for an example multi-destination campaign.
Figure 35: Scenario feasibility visualization. Cumulative raw and remaining transport capacity
must always exceed estimated cumulative demands for feasible campaigns.
Supply Network Visualization
The supply network visualization displays transports and aggregated demands in a time-expanded
network and is often a precursor to the manifesting process. Line width represents transport capacity,
transport demands, or net transport capacity (the difference between the two). Circle diameter represents
the mass of aggregated exploration period demands.
61
Figure 36: Supply network visualization. Line width represents net transport capacity and
circle diameter represents aggregated exploration period demand mass.
Demand History Visualizations
In addition to file-format export options, charts filtered by elements, locations, or missions visualize raw
demands. Demand history by element provides a time-history of the demands generated by a single
element. In some cases, such as mission demand models, demands for resources are not generated by a
particular element. Demand history by location provides a time-history of demands generated at a node or
edge. Finally, demand history by mission provides a time-history of the demands generated by any
element or location during a mission. In all three cases, demands are grouped by base class of supply.
Figure 37: Element demand history visualization. Plots demands for spares (COS 4) and
packing mass estimates (COS 5) generated by a habitat element. The non-linear response results
from reconfigurations of element state during crewed and un-crewed periods.
62
Commonality Analysis
The commonality analysis tab provides details of campaign part commonality and insights to how
scavenging of parts impacts demands generation. If enabled, scavenging events are logged by the demand
simulator and displayed with filters by the source of scavenged parts and the type of parts scavenged.
Figure 38: Commonality and scavenging analysis. Displays scavenged common parts filtered
by source element and type of resource.
Repairability Analysis
The repairability analysis tab establishes a repair list for each crewed mission in an exploration campaign.
Currently, repair activities are limited to crewed missions with deterministic demands; future research is
required for extensions to more detailed campaigns. To efficiently use the available time, a repair policy
should maximize the benefit (mass savings) for a given cost (crew time). In addition to manually-selected
repair, an “Auto-repair” option chooses an optimal set of items to repair.
The repairable items demanded in a mission may be considered to be in random order. Only items
demanded during a crewed mission or from non-crewed missions preceding it are considered. Each item i
is associated with a mass benefit, the unit mass (Mi) less the mass to repair (MTRi), and a time cost, the
time to repair (TTRi). If the repairable items are sorted in descending order of the benefit-cost ratio
iii TTRMTRM )( the set of items to repair to optimize the mass savings can be chosen up to a crew time
availability. Figure 39 illustrates a graphical representation of this principle. This practice is repeated for
each crewed mission where repair activities are allowed. Non-optimal repair lists will fall below the
sorted repair list curve, indicating sub-optimal mass savings for a fixed repair time.
63
Figure 39: Repair policy for optimal mass savings. The set of items to repair to achieve optimal
mass savings is found by sorting the repairable items in descending order by the benefit-cost ratio.
4.3.5 Manifest Tab
The manifest tab assists building a cargo manifest, following the steps of the manifesting model: demand
generation, aggregation, packing, and manifesting (see Section 3.3). Manual manifesting consists of
sequentially packing demands into resource containers and manifesting the containers onto carriers.
Figure 40: Packing and manifesting interface. The manifesting tab helps the user to create a
valid cargo manifest by guiding through the process of packing and manifesting containers.
64
Although demand generation and aggregation are automatic, packing resources into containers is the first
task of manifesting. Since containers may be transferred across several transports before the resources are
demanded, it is challenging to share containers across multiple spatial and temporal demand periods. The
simplest packing scheme only allows demands from the same transport or exploration period to share a
container, however there are some simple extensions that can be useful for sharing containers. In general,
a partially-filled resource container may only be assigned more cargo if it has not reached its target
destination before the new resources are demanded. Transport demands can be packed within a partially-
filled container if it contains demands from the same transport or subsequent exploration periods at the
transport destination node. Exploration demands can be packed within a partially-filled container if it
contains demands from the same or later explorations at the same node.
After packing is complete, manifesting containers onto carriers must take place. Resource containers
containing exploration demands may be manifested on one of the carriers of any transport arriving at the
location of the exploration location at the same or earlier time as the earliest packed demand exploration
period. However, if the origin of the transport is not a source node, the container must be supplied to the
transport in a derivative demand.
Manual manifesting is tedious due to the recursive process of manifesting containers on subsequent
transports. To help automate the process, a heuristic algorithm was implemented to create manifests. The
resulting manifests are not optimal, meaning some feasible campaigns may appear infeasible and human
modification may be necessary to correct logical errors.
The auto-manifesting algorithm depends on an auto-packing routine, shown below in pseudo code. The
auto-packing scheme selectively uses existing partially-filled containers before creating new containers.
procedure auto-pack (Demand D) {
for each existing resource container C {
if D can be packed in C {
pack D in C until capacity is reached
}
}
while remaining D exists {
create new resource container C
pack D in C until capacity is reached
}
}
The logic for the auto-manifesting algorithm puts emphasis on carry-along supplies, meaning demands
are preferentially manifested on the transport closest in time before the point of demand, outlined below
in pseudo code. Auto-packing is performed before manifesting each exploration period in attempt to
65
promote sharing of partially-packed containers. Also, priority is given to transport demands, as they must
be manifested on the transport from which they are demanded.
procedure auto-manifest () {
for each exploration period E, in reverse chronological order {
T is the transport that supplies E
for each demand D aggregated to T {
auto-pack (D)
}
for each demand D aggregated to E {
auto-pack (D)
}
for each resource container C containing a demand from T {
for each transport S supplying E, in reverse chron. order {
for each carrier R in S {
if C can be manifested onto R {
manifest C onto R
go to next container
}
}
}
C could not be manifested: demands may be infeasible
}
for each container C containing a demand from E {
for each transport S supplying E, in reverse chron. order {
for each carrier R in S {
if C can be manifested onto R {
manifest C onto R
go to next container
}
}
}
C could not be manifested: demands may be infeasible
}
}
}
The known limitations of auto-manifesting include the effects of carriers involved in multiple transports.
First, carrier capacity calculations don‟t take into account containers manifested on previous transports
not yet transferred to another carrier. This effect is partially due reverse-chronological manifesting and
can cause spatial simulation errors if the auto-manifested containers violate mass or volume capacity
limits. Second, containers are only transferred to a carrier immediately before a transport, potentially
allowing containers to be inadvertently moved by carriers after delivery to a location. Future development
efforts, in particular those of optimal manifesting methods discussed in Appendix B, should correct the
deficiencies of present auto-manifesting.
4.3.6 Simulation Tab
The final simulation and analysis is performed with the simulation tab. In addition to simulation outputs
including errors, it provides several visualizations for campaign analysis.
66
Network State History
The network state history visualization provides an animation of the movement of elements throughout
the network. The animation runs forward at a set frame rate with either fixed simulation time per frame or
one event per frame. Selecting specific dates enable investigation of the network state including element
locations, resource amounts, and other element properties.
Figure 41: Network state history. The network state history shows the system state after each
event execution. Elements can be inspected for contents including remaining propellant mass.
Measures of Effectiveness History
The measures of effectiveness (see Section 3.4.2) are the primary quantitative outputs of campaign
simulation to assist analysis. Rather than simply providing a number to quantify the campaign, however,
measure of effectiveness visualizations provide a time-history of metrics‟ evolution. The visualizations
may help identify parts of a campaign that could be
changed to provide improvement in a particular
metric. Figure 42 illustrates an example exploration
campaign in which the relative exploration capability
(REC) does not increase monotonically as expected
due to non-exploration mass launched.
Resource History Visualizations
Resource history visualizations track the supply and
demand of resources during the campaign. Resources Figure 42: Relative exploration capability
history. REC does not increase monotonically over
a campaign due to non-exploration mass launched.
67
are grouped by base class of supply and can be filtered by network location or by element. Resource
history by element is best used to track propellant usage during burns and resource history by location is
best used to track exploration consumption.
Figure 43: Location resource history. The resource history during an example campaign shows
a build-up of COS 5, representing empty resource containers.
68
5 SpaceNet 2.5 Applications
This chapter presents a series of four case studies and a controlled user study to illustrate the applications
of SpaceNet 2.5 in space exploration campaign analysis. The case studies are targeted towards campaigns
considered for future human space exploration to demonstrate feasibility and flexibility of the SpaceNet
modeling framework. The user study is focused on highlighting the usability and timescales required for
performing campaign analysis using the SpaceNet 2.5 software.
5.1 Case Studies
The four case studies selected represent campaigns considered during the development of SpaceNet 2.5
and new concepts under development. First, as an example of an operational space exploration campaign
with many flights, the resupply of the International Space Station is considered between 2010 and 2015.
Second, a sortie exploration to a near-Earth object (NEO) demonstrates an exploration scenario not
considered during initial development. Third, a lunar outpost exploration campaign similar to NASA
Constellation program plans is used as an example of a long-duration planetary exploration with
significant surface infrastructure. Finally, a Mars surface exploration similar to a NASA design reference
mission is used as an example of an exploration having long-duration transportation segments.
As demonstrative case studies, the level of fidelity and technical correctness and completeness is limited –
a more in-depth analysis for each case should be undertaken to establish validated results. In particular,
the most visible inconsistencies may be present in the ISS resupply case study as it attempts to analyze an
existing system using an imperfect model and limited information. As conceptual campaigns, the other
three case studies model technologies and designs under development, requiring refinements as unknowns
are resolved.
5.1.1 International Space Station Resupply
At the time of writing, there remain two scheduled flights of the NASA Space Transportation System
(STS) while the ISS lifetime has been extended to 2020 or beyond [20]. Maintaining the crew and
operations at the ISS in the coming years without the support of the shuttle is not a trivial task [34].
NASA has indicated commercial on-orbit transportation services (COTS), also known as commercial
resupply services (CRS), will play a large role in supplying the ISS by issuing contracts to Orbital and
SpaceX for use of Cygnus and Dragon spacecraft respectively [35]. Combined with the efforts of ESA‟s
69
automated transfer vehicle, ATV, JAXA‟s H-II transfer
vehicle, HTV, and RKA‟s Progress and Soyuz, the ISS
will become a complicated supply hub.
The goal of this case study is to model the final
assembly and subsequent resupply of the ISS including
all scheduled and expected flights through December
2015. Without sophisticated demand models to estimate
demands for individual resources and spares, the
analysis will focus on lumped mass demands by class of
supply using parametric models for crew consumables
and spares [36]. The case study does not consider down-
mass capability, improvements to launch vehicles or
spacecraft capacities, resources pre-positioned at the
ISS before 2010, differences between cargo types (e.g. dry, water, or gas), or individual crew rotations.
Model Inputs
Most model inputs for the ISS resupply case study were derived from spacecraft datasheets where
available or from publicly-available online databases. All values are approximate due to modeling
simplifications and assumptions, vehicle configurations, and design evolution.
Table 19 lists the nodes considered in this case including launch and landing sites on Earth and the ISS in
orbit. By modeling the entire ISS as a single node, resources are shared between all modules in orbit.
Table 19: ISS resupply nodes.
Abbrev. Description Parameters
ISS International Space Station 360 km x 347 km, 51.6°
KSC Kennedy Space Center 28.6° N, 80.6° W
CCAS Cape Canaveral Air Station 34.9° N, 117.8° W
WFF Wallops Flight Facility 28.4° N, 80.6° W
TSC Tanegashima Space Center 30.4° N, 131.0° E
GSC Guiana Space Center 5.0° N, 52.8° W
BCD Baikonur Cosmodrome 45.9° N, 63.3° E
SLZ Soyuz Landing Zone 50° N, 67.5° E
PSZ Pacific Splashdown Zone 15° N, 160° W
Figure 44: ISS resupply network.
Visualization of Earth launch sites, landing
zones, and the ISS in orbit.
70
Table 20 lists the edges corresponding to launch vehicle capabilities and landings. Using abstracted flight
edges avoids the definition of launch vehicles and propellants. In all cases but the STS, launch vehicles
are independent from the crew and cargo carriers, enabling for a clean definition of the flight edges. In the
case of the STS, the flight edge represents the carrying capacity of the shuttle without its structural mass.
Table 20: ISS resupply edges.
Name Origin Destination Max Crew Max Cargo
[kg]
Soyuz-FG Launch to ISS BCD ISS 3 7200
Soyuz-2 Launch to ISS BCD ISS 3 8500
Soyuz Landing at SLZ ISS SLZ 3 150
Falcon 9 Launch to ISS CCAS ISS 0 10450
Dragon Splashdown at PSZ ISS PSZ 0 3000
Taurus II Launch to ISS WFF ISS 0 6600
H-IIB Launch to ISS TSC ISS 0 16500
Ariane 5 ES Launch to ISS GSC ISS 0 19300
STS Launch to ISS KSC ISS 7 16050
STS Shuttle Landing at KSC ISS KSC 10 9500
Proton-M Launch to ISS BCD ISS 0 21600
Table 21 lists the element definitions used in the case study. Many elements correspond to the spacecraft
carrying crew and cargo to the ISS, though some represent infrastructure and logistics containers. As
discussed with the flight edges, the STS shuttle element does not include its infrastructure mass which is
considered in the flight edge parameters.
Both the ISS and its crew produce demands with linear demand models. Annual ISS demands, including
packaging mass, are estimated at 10 tons of spares and maintenance and 15 tons of science payloads.
Daily crew demands are estimated at 2 kilograms of food, 3.5 kilograms of water, 1 kilogram of gases,
0.5 kilogram of hygiene items, and 0.5 kilogram of waste disposal items per crew member.
71
Table 21: ISS resupply elements.
Name Dry Mass
[kg] Max Crew
Max Cargo
[kg] Description
Progress-M 4900 0 2350 RKA Progress (M Configuration)
Soyuz-TMA 6085 3 100 RKA Soyuz (TMA Configuration)
Dragon 4200 0 6000 SpaceX Dragon
Cygnus 3500 0 2000 Orbital Cygnus
Cygnus-M 3500 0 2700 Orbital Cygnus (Improved)
HTV 8100 0 6000 JAXA H-II Transfer Vehicle
ATV 11700 0 7600 ESA Automated Transfer Vehicle
STS Shuttle 0 10 16050 NASA Space Transportation System Shuttle
MLM 20300 0 0 RKA Multifunctional Laboratory Module
ELC 4400 0 2000 EXPRESS Logistics Carrier
PMM 4080 0 9070 Pressurized Multipurpose Module
AMS 6700 - - Atomic Magnetic Spectrometer
ISS 335000 6 35000 International Space Station
Missions
The official mission manifest provided by NASA only covers missions through the end of 2010 [36]. A
complete mission manifest through 2015 was created using unofficial launch and mission manifests
provided by Orbital, SpaceX, JAXA, and ESA, as well as extrapolating launch rates for Progress and
Soyuz. The missions are comprised of 2 STS, 22 Progress, 22 Soyuz, 12 Dragon, 8 Cygnus, 5 HTV, and 4
ATV resupply missions and one assembly mission to replace the Pirs module with Nauka. In addition to
the resupply missions, the first mission, number zero, initializes the ISS and its crew in orbit to start the
demands generation.
Although it is immaterial to this analysis, it is assumed that each Soyuz spacecraft spends 180 days
docked at the ISS before the subsequent return to Earth. All other spacecraft spend 60 days docked at the
ISS before de-orbiting or return to Earth.
72
Table 22: ISS resupply missions 0-22 (2010-2011).
# Date Mission Flight(s) Element(s)
0 9/1/2010 (Initial Conditions) - ISS, Crew Members A-F
1 9/8/2010 Progress M-08M Soyuz-FG Launch Progress M “M-08M”
2 9/16/2010 STS 133 STS Launch
STS Shuttle Landing
STS Shuttle “Discovery”
ELC “ELC3”
PMM “Leonardo”
3 9/29/2010 Soyuz TMA-01M Soyuz-FG Launch
Soyuz Landing
Soyuz-TMA “TMA-01M”
4 10/27/2010 Progress M-09M Soyuz-FG Launch Progress M “M-09M”
5 11/10/2010 STS 134 STS Launch
STS Shuttle Landing
STS Shuttle “Endeavor”
ELC “ELC4”
AMS “AMS-02”
6 11/18/2010 ATV-2 Ariane 5 ES Launch ATV “Johannes Kepler”
7 11/30/2010 Soyuz TMA-20 Soyuz-FG Launch
Soyuz Landing
Soyuz-TMA “TMA-20”
8 1/1/2011 HTV-2 H-IIB Launch HTV “HTV-2”
9 2/9/2011 Progress M-10M Soyuz-FG Launch Progress M “M-10M”
10 3/1/2011 Soyuz TMA-21 Soyuz-FG Launch
Soyuz Landing
Soyuz TMA “TMA-21”
11 3/15/2011 Progress M-11M Soyuz-FG Launch Progress M “M-11M”
12 4/1/2011 HTV-3 H-IIB Launch HTV “HTV-3”
13 4/15/2011 Progress M-12M Soyuz-FG Launch Progress M “M-12M”
14 5/1/2011 Dragon-1 Falcon 9 Launch
Dragon Splashdown
Dragon “Dragon-1”
15 5/15/2011 Progress M-13M Soyuz-FG Launch Progress M “M-13M”
16 5/30/2011 Soyuz TMA-02M Soyuz-FG Launch
Soyuz Landing
Soyuz-TMA “TMA-02M”
17 6/15/2011 Progress M-14M Soyuz-FG Launch Progress M “M-14M”
18 7/15/2011 Progress M1-01M Soyuz-2 Launch Progress M “M1-01M”
19 8/1/ 2011 Cygnus-1 Taurus II Launch Cygnus “Cygnus-1”
20 8/15/2011 Progress M-15M Soyuz-FG Launch Progress M “M-15M”
21 9/15/2011 Soyuz TMA-22 Soyuz-FG Launch
Soyuz Landing
Soyuz TMA “TMA-22”
22 11/1/2011 Dragon-2 Falcon 9 Launch
Dragon Splashdown
Dragon “Dragon-2”
73
Table 23: ISS resupply missions 23-45 (2011-2013).
# Date Mission Flight(s) Element(s)
23 12/1/2011 Soyuz TMA-03M Soyuz-FG Launch
Soyuz Landing
Soyuz-TMA “TMA-03M”
24 12/15/2011 Progress M-16M Soyuz-FG Launch Progress M “M-16M”
25 12/30/2011 Nauka Assembly Proton-M Launch MLM “Nauka”
26 2/1/2012 Progress M-17M Soyuz-FG Launch Progress M “M-17M”
27 3/1/2012 Soyuz TMA-04M Soyuz-FG Launch
Soyuz Landing
Soyuz-TMA “TMA-04M”
28 4/1/2012 HTV-4 H-IIB Launch HTV “HTV-4”
29 4/15/2012 ATV-3 Ariane 5 ES Launch ATV “Edoardo Amaldi”
30 5/1/2012 Dragon-3 Falcon 9 Launch
Dragon Splashdown
Dragon “Dragon-3”
31 5/15/2012 Soyuz TMA-05M Soyuz-FG Launch
Soyuz Landing
Soyuz-TMA “TMA-05M”
32 7/1/2012 Progress M-18M Soyuz-FG Launch Progress M “M-18M”
33 8/1/2012 Soyuz TMA-06M Soyuz-FG Launch
Soyuz Landing
Soyuz-TMA “TMA-06M”
34 9/1/2012 Cygnus-2 Taurus II Launch Cygnus “Cygnus-2”
35 10/1/2012 Dragon-4 Falcon 9 Launch
Dragon Splashdown
Dragon “Dragon-4”
36 11/1/2012 Soyuz TMA-07M Soyuz-FG Launch
Soyuz Landing
Soyuz-TMA “TMA-07M”
37 12/1/2012 Progress M-19M Soyuz-FG Launch Progress M “M-19M”
38 1/1/2013 Soyuz TMA-08M Soyuz-FG Launch
Soyuz Landing
Soyuz-TMA “TMA-08M”
39 2/1/2013 Progress M-20M Soyuz-FG Launch Progress M “M-20M”
40 3/1/2013 Dragon-5 Falcon 9 Launch
Dragon Splashdown
Dragon “Dragon-5”
41 4/1/2013 HTV-5 H-IIB Launch HTV “HTV-5”
42 5/1/2013 Cygnus-3 Taurus II Launch Cygnus-M “Cygnus-3”
43 6/1/2013 Soyuz TMA-09M Soyuz-FG Launch
Soyuz Landing
Soyuz-TMA “TMA-09M”
44 7/1/2013 Dragon-6 Falcon 9 Launch
Dragon Splashdown
Dragon “Dragon-6”
45 8/1/2013 Soyuz TMA-10M Soyuz-FG Launch
Soyuz Landing
Soyuz-TMA “TMA-10M”
74
Table 24: 2014 ISS resupply missions 46-68 (2013-2015).
# Date Mission Flight(s) Element(s)
46 9/1/2013 Progress M-21M Soyuz-FG Launch Progress M “M-21M”
47 9/15/2013 ATV-4 Ariane 5 ES Launch ATV “ATV-004”
48 10/1/2013 Cygnus-4 Taurus II Launch Cygnus-M “Cygnus-4”
49 11/1/2013 Dragon-7 Falcon 9 Launch
Dragon Splashdown
Dragon “Dragon-7”
50 11/15/2013 Soyuz TMA-11M Soyuz-FG Launch
Soyuz Landing
Soyuz-TMA “TMA-11M”
51 12/15/2013 Progress M-22M Soyuz-FG Launch Progress M “M-22M”
52 1/15/2014 Soyuz TMA-12M Soyuz-FG Launch
Soyuz Landing
Soyuz-TMA “TMA-12M”
53 2/15/2014 Progress M-23M Soyuz-FG Launch Progress M “M-23M”
54 3/1/2014 Dragon-8 Falcon 9 Launch
Dragon Splashdown
Dragon “Dragon-8”
55 4/1/2014 HTV-6 H-IIB Launch HTV “HTV-6”
56 4/15/2014 Cygnus-5 Taurus II Launch Cygnus-M “Cygnus-5”
57 5/1/2014 Soyuz TMA-13M Soyuz-FG Launch
Soyuz Landing
Soyuz-TMA “TMA-13M”
58 6/1/2014 Dragon-9 Falcon 9 Launch
Dragon Splashdown
Dragon “Dragon-9”
59 7/1/2014 Progress M-24M Soyuz-FG Launch Progress M “M-24M”
60 8/1/2014 Soyuz TMA-14M Soyuz-FG Launch
Soyuz Landing
Soyuz-TMA “TMA-14M”
61 9/1/2014 Dragon-10 Falcon 9 Launch
Dragon Splashdown
Dragon “Dragon-10”
62 10/1/2014 Progress M-25M Soyuz-FG Launch Progress M “M-25M”
63 11/1/2014 Soyuz TMA-15M Soyuz-FG Launch
Soyuz Landing
Soyuz-TMA “TMA-15M”
64 12/1/2014 Cygnus-6 Taurus II Launch Cygnus-M “Cygnus-6”
65 1/1/2015 Progress M-26M Soyuz-FG Launch Progress M “M-26M”
66 2/1/2015 Soyuz TMA-16M Soyuz-FG Launch
Soyuz Landing
Soyuz-TMA “TMA-16M”
67 2/15/2015 ATV-5 Ariane 5 ES Launch ATV “ATV-005”
68 3/1/2015 Dragon-11 Falcon 9 Launch
Dragon Splashdown
Dragon “Dragon-11”
75
Table 25: ISS resupply missions 69-77 (2015).
# Date Mission Flight(s) Element(s)
69 4/1/2015 HTV-7 H-IIB Launch HTV “HTV-7”
70 5/1/2015 Soyuz TMA-17M Soyuz-FG Launch
Soyuz Landing
Soyuz-TMA “TMA-17M”
71 6/1/2015 Cygnus-7 Taurus II Launch Cygnus-M “Cygnus-7”
72 7/1/2015 Progress M-27M Soyuz-FG Launch Progress M “M-27M”
73 8/1/2015 Soyuz TMA-18M Soyuz-FG Launch
Soyuz Landing
Soyuz-TMA “TMA-18M”
74 9/1/2015 Dragon-12 Falcon 9 Launch
Dragon Splashdown
Dragon “Dragon-12”
75 10/1/2015 Cygnus-8 Taurus II Launch Cygnus-M “Cygnus-8”
76 11/1/2015 Progress M-28M Soyuz-FG Launch Progress M “M-28M”
77 12/1/2015 Soyuz TMA-19M Soyuz-FG Launch
Soyuz Landing
Soyuz-TMA “TMA-19M”
Analysis and Discussion
The mission bat chart, shown in Figure 45, highlights the immense number of transports to and from ISS
between August 2010 and December 2015. Docking activities at the ISS will require action for 18 arrivals
per year on average, or one arrival every 20 days. For comparison, there were 14 transports to ISS in
2009, and 11 in 2008, and only 9 in 2007.
Figure 45: ISS resupply mission bat chart. Green squares are element instantiations, yellow
lines are flight transports, black squares are elements removed from simulation.
76
Due to the simplifications in modeling demands, analysis of the ISS resupply case is limited to aggregate
logistics feasibility rather than an intensive cargo manifesting analysis. The most useful visualization is
the scenario feasibility plot, shown in Figure 46. Since this case only has only a single destination, it is
feasible as the estimated demands remain below the remaining capacity line. The total raw capacity to ISS
over the simulation is 245 tons, with 225 tons remaining after considering delivered elements on the
shuttle missions (ELC, AMS, etc.). The demands over the same time period total 220 tons, of which 82
tons are for crew provisional items (COS 2), 80 tons are payloads for exploration and research (COS 6),
52 tons are for maintenance and upkeep (COS 4), and 6 tons are for waste and disposal (COS 7).
Figure 46: ISS resupply scenario feasibility. Cumulative plot of raw supply vehicle capacity,
remaining capacity after pre-manifested elements, and aggregated demands indicating feasibility.
Although not modeled, any pre-positioned resources at ISS would effectively shift the estimated demands
curve down by a fixed amount no more than the maximum estimated capacity of 35 tons. Analysis
without considering these pre-positioned resources focuses on the steady-state supply and demand. The
steady-state supply margin repeatedly falls below 5 tons, warranting additional analyses using two
strategies. First, sensitivity studies for launch schedule and spacecraft availability help identify periods of
high risk for supply. Second, sensitivity studies for demand rates help illustrate the effects of technology
improvement on the reduction of demands.
As a hypothetical example of the impact of spacecraft availability, consider the conceivable case that one
of the COTS vehicles fail to meet their contract. In Figure 47, all of the Orbital Cygnus spacecraft have
77
been removed from the supply capacity lines. The first signs of infeasibilities appear in early 2013 in this
scenario, however depending on the amount of pre-positioned resources not modeled, there may not be
any serious problems until 2014 or 2015, where the infeasible margin exceeds 10 tons.
Figure 47: ISS resupply feasibility with Cygnus COTS failure. Without the Cygnus spacecraft,
steady-state infeasibilities fist start to appear in early 2013 and only worsen into 2014 and 2015.
As another hypothetical example, consider a 5-ton advanced water recovery system (AWRS) delivered by
Dragon-6 in July 2013 capable of reducing the crew water demands from 3.5 to 0.5 kilograms per person
per day. This trade between delivery capacity and future demands is modeled by the adding a new
element, the AWRS, to the Dragon-6 before launch. After delivery to the ISS, the six crew members are
reconfigured to a new state representing the lower demands for water resources.
Figure 48 shows the resulting feasibility chart highlighting the benefits that could be realized with such a
decision. Although the cumulative remaining capacity is reduced by 5 tons, the mass of the AWRS, the
cumulative demands decrease by a much more substantial margin of almost 40 tons by the end of 2015.
Although only a notional example, similar analysis could inform technology development for advanced
life support systems.
78
Figure 48: ISS resupply with advanced water recovery. An advanced water recovery system is
delivered in July 2013 reducing demands for water from 3.5 to 0.5 kg/person/day.
Conclusions
Though only a high-level analysis without access to detailed demand models, the resupply of the ISS
through 2015 warrants significant additional research. As modeled, there is limited supply capacity in
steady-state, indicating undersupplies of critical resources may be a realistic concern. Steady-state
infeasibilities could start to occur with the delay or cancellation of just one of the six resupply spacecraft.
Also, as there is no net aggregation in resources or a high-capacity resupply vehicle, additional
infeasibilities could come from a multiple failure event depleting any available spares, a conceivable
possibility with the next solar maximum occurring in 2013. In order to reduce demands, efforts should
also be taken to implement high-closure ECLSS systems as early as possible for maximum impact.
More detailed analysis should include additional demands for propellant required for orbital re-boost and
station keeping and differentiate between pressurized, unpressurized and liquid cargo, including the
multiple spacecraft configurations supporting differing capacities of each type. Provided additional
information, pre-positioned resources could be modeled aboard ISS to provide a more sophisticated
feasibility analysis. Finally, down-mass capacity should be quantified to inform capability to return failed
equipment for repair and analysis.
79
5.1.2 Near-Earth Object Sortie
Concepts for human explorations to the asteroids, comets, and other objects having similar orbits to Earth
(collectively called near-Earth objects, or NEOs) have existed as early as 1966 when Smith proposed a
500-plus day mission to Eros using modified Apollo spacecraft [38]. Such a mission could gain
operational experience outside the Earth-Moon system without the cost required for a Martian landing.
More recently, NASA‟s Advanced Projects Office performed feasibility studies for missions to NEOs
[39, 40]. NEO exploration could improve technical readiness levels for space hardware, evaluate in-situ
resource utilization systems, and provide a wealth of new information. Also, exploration would resemble
docking operations and EVAs without requiring landing or ascent spacecraft. The NEOs investigated are
reachable with 150-day missions including 1999 AO10, 2000 SG344, and 2006 DQ14.
The HSF review committee‟s final report also mentions
exploration of NEOs as one step on a flexible path to
human space exploration [18]. The Flexible Path
strategy calls for explorations of increasing duration and
technical difficulty, starting with a lunar flyby or orbit,
visiting Lagrange points, and exploring NEOs before
exploring Mars.
The feasibility of a 14-day exploration at NEO 1999
AO10 is considered for this case study due to a
favorable launch opportunity within a conceivable timeline. Past research claims 1999 AO10 is reachable
using Constellation program spacecraft with limited modifications including the Orion crew exploration
vehicle and Ares V heavy lift launch vehicle [40]. The target of analysis is to evaluate the feasibility of
such a mission and to propose modifications if required.
Figure 50: NEO sortie network. Visualization of space edges between launch, low-Earth orbit,
the near-Earth object 1999 AO10, and splashdown in the Pacific Ocean.
Figure 49: Flexible path strategy. Timeline of
milestones, destinations and capabilities of the
Flexible Path strategy, adapted from [18].
80
Model Inputs
The nodes required for this campaign are limited to the launch and landing sites on Earth, the parking
orbit in low-Earth orbit, and 1999-AO10. The NEO is roughly modeled as an orbital node about the sun.
Table 26: NEO sortie nodes.
Abbrev. Description Parameters
KSC Kennedy Space Center 28.6° N, 80.6° W
LEO Low Earth Parking Orbit 166.7 km x 166.7 km, 38.0°
AO10 Near-Earth Object 1999 AO10 1.01 AU x 0.81 AU, 2.62°
PSZ Pacific Splashdown Zone 15° N, 160° W
The edges represent launch and in-space propulsive burn sequences based on delta-v estimates for a
mission departing in September 2025 [40]. Reaction control system (RCS) and attitude control burns are
not considered.
Table 27: NEO sortie edges.
Name Origin Destination Duration
[days]
Propulsive Burns
KSC Launch to LEO KSC LEO 1 t+0.0: 9.8 km/s (OMS)
LEO to AO10 LEO AO10 111 t+0.0: 3.291 km/s (OMS)
t+111: 2.193 km/s (OMS)
AO10 to Splashdown at PSZ ISS SLZ 31 t+0.0: 1.746 km/s (OMS)
All element designs are notional as no current plans exist for a NEO mission, but based on available data
for Constellation program spacecraft. The elements used for the mission include an Ares V heavy-lift
launch vehicle and an Orion crew exploration vehicle. The Orion modifications include reducing the crew
module (CM) crew size to two to improve the cargo storage capacity and the inclusion of a scientific
instrumentation payload (SIP) placed inside the service module (SM).
A few assumptions are included for in-space propulsion. First, zero-loss cryo-coolers are assumed to be
available to prevent the boil-off of the EDS cryogenic propellant over the duration of the transit to 1999-
AO10. Also, both the EDS and SM are assumed to be restartable. Additionally, as the crew exploration
vehicle does not contain an airlock, all exploration will be tele-operated without EVAs.
To model demands, each crew member is assigned a linear demand model to generate 7.5 kilograms per
day of generic crew consumables (COS 2). All required spares and maintenance mass for the mission
duration is assumed to be included in element dry mass estimates.
81
Table 28: NEO sortie elements.
Name Dry Mass
[kg] Max Crew
Max Cargo
[kg]
Fuel [kg]
(Type) Isp [s] Description
SRBs 213000 0 0 1,370,000
PBAN
269 Solid Rocket Boosters (2)
Core 173,680 0 0 1,587,000
LOX/LH2
414 Core Stage
Interstage 9190 - - - - Interstage Element
EDS 26,390 0 0 253,000
LOX/LH2
449 Earth Departure Stage
CM 9600 2 1500 - - Crew Module
SM 3000 0 1000 10,000
MMH/N2O4
301 Service Module
SIP 1000 - - - - Science Instrument Payload
LAS 3700 0 0 2500
HTPB
250 Launch Abort System
SA 500 - - - - Spacecraft Adapter
Missions
There is only one sortie mission to 1999-AO10. The launch from Kennedy Space Center uses a staging of
the two solid rocket boosters, the Ares V core, and the Earth departure system to achieve low-Earth orbit.
Once in low-Earth orbit, the Earth departure stage is used to depart from Earth orbit. Upon arrival at 1999
AO10, the Earth departure stage is fired for the last time followed by a burn of the service module. The
exploration operations take place over a period of two weeks, followed by a return to Earth using the
service module engine.
Table 29: NEO sortie mission events.
Date(s) Event Details
9/19/2025-
9/19/2025
Launch from KSC to LEO
parking orbit
Burn Boosters, Stage Boosters,
Stage LAS, Burn Core, Stage Core,
Stage Interstage, Burn EDS
9/19/2025-
1/08/2026
Traverse in-space edge to
1999 AO10
(Departure) Burn EDS
(Arrival) Burn EDS, Stage EDS,
Stage SA, Burn SM
1/08/2026-
1/22/2026
Exploration at AO10 Tele-operated exploration (no EVAs)
1/22/2026-
2/22/2026
Traverse in-space edge to
Pacific splashdown
Burn SM, Stage SM
82
Analysis and Discussion
As initially specified, the mission to 1999 AO10 is infeasible due to insufficient propellant for the return
transport to Earth. Even without considering manifested cargo to satisfy the crew demands, there is
approximately 525 m/s delta-v unachievable by the service module burn. In addition, the mission is
logistically infeasible because there is not enough cargo space aboard the crew module to satisfy the
demands generated over the course of the 150-day mission. In fact, the transit from low-Earth orbit to
1999 AO10 itself exhausts the 1500 kilogram cargo capacity without considering packing mass.
a)
b)
Figure 51: NEO sortie infeasibilities. (a) The service module burn to return to Earth has a 526.8 m/s delta-v deficit
without considering cargo. (b) The 111-day transit to 1999-AO10 alone exhausts the crew module cargo capacity.
In attempt to correct these infeasibilities, a modified NEO sortie was constructed with several changes to
improve the performance. The included changes may not be technologically possible, but are presented as
a method to iterate on infeasible mission designs.
Increase EDS fuel capacity by 20% from 253,000 to 303,600 kilograms
Decrease EDS mass by 10% from 26,390 to 23,750 kilograms
Increase CM cargo capacity from 1500 to 2500 kilograms
Decrease CM mass by 10% from 9600 to 8640 kilograms
Increase SM fuel capacity by 10% from 10,000 to 11,000 kilograms
Science instrumentation payload is not returned to Earth
The revised NEO exploration is both propulsively and logistically feasible, though with tight propellant
margins. Although the delta-v calculation does take into account the mass of manifested resources, all
83
resources are assumed to be discarded from the spacecraft after consumption. The service module only
has 250 kilograms of excess propellant upon its return to Earth, as shown in Figure 52.
Figure 52: Modified NEO sortie service module resource history. The science instrument
payload (COS 6) is discarded. The service module propellant (COS 1) margin is 250 kg.
A similar inspection of the crew module in Figure 53 highlights the consumption of the crew consumable
resources as well as the usage of the packing materials. In this case, the automated packing method
exclusively uses cargo transfer bags (CTBs) and half CTBs with nearly a 100% packing fraction for each
container due to the continuous nature of the abstracted resources. Additional analysis may restrict the
volume of resources in each CTB or introduce less mass-efficient containers such as gas and water tanks.
Figure 53: Modified NEO sortie crew module resource history. All crew consumables (COS 2)
are used in the crew module. Packing materials (COS 5) total about 100 kg of the cargo.
84
There are two residual concerns about the feasibility of the mission as stated. First, there are no provisions
for an airlock on the crew module, requiring all crewed exploration would take place from within the
spacecraft. This would still provide scientific value, as crew could tele-operate rovers or other robotic
elements on the surface of the asteroid, but it would not be as fulfilling as an EVA. The addition of an
airlock to the spacecraft would significantly increase the mass, further constraining feasibility.
The second concern is for the boil-off of the cryogenic propellants used by the EDS. As planned for the
lunar exploration, the EDS would only be used over the course of a few days to a week, limiting the time
for the liquid oxygen and liquid hydrogen to boil-off. In the case of an exploration to 1999 AO10, the
time between launch and arrival would surpass 100 days, significantly impacting the availability of the
propellants. Further research should be undertaken to identify the boil-off rates for cryogenic propellants
over long durations, including identifying enabling technologies to make this mission feasible.
Conclusions
A sortie mission to a near-Earth object such as 1999 AO10 is not feasible without significant
modifications to existing Constellation program spacecraft to reduce mass and increase fuel capacity.
Aside from concerns with the storage of cryogenic propellants for the 100-plus day transit to the asteroid,
the existing element designs do not provide sufficient storage capacity for the additional service module
propellant, the crew consumable supplies, packaging and tare mass for supplies, and the scientific
equipment needed to perform studies once at the asteroid. With additions such as an expanded in-space
habitat for the crew, a dedicated in-space propulsion stage, an airlock to perform EVAs, and a third crew
member to assist with EVAs, the mission calls for a different set of spacecraft more in-line with a Martian
rather than a lunar exploration.
5.1.3 Lunar Outpost Buildup
An extended lunar exploration building to continuous human presence was one potential goal of NASA‟s
Constellation program. Although the fate of the program is currently uncertain, it still serves as an
adequate case study as an example of an exploration campaign with significant use of surface operations.
In addition, due to the maturity of the campaign architecture, the modeled exploration benefits from
detailed and realistic element models based on data developed with a reasonable amount of analysis.
As of late 2009, the most current iteration of the lunar surface architecture was Scenario 12, developed by
the NASA Lunar Surface Systems Project Office (LSSPO) and the Constellation Architecture Team –
Lunar (CxAT-Lunar). Scenario 12 has evolved from the confluence of three previous scenarios: Scenario
4 (Optimized Exploration), Scenario 5 (Fission Surface Power System), and Scenario 8 (Initial Extensive
85
Mobility) [41, 42]. In Scenario 12, successive missions at a
rate of about three per year deliver infrastructure components
to an outpost, building up to full capability within six years.
One of the interesting aspects of Scenario 12 from an
analysis perspective is a high degree of reuse of surface
mobility elements for excursions and general exploration.
The lunar electric rover (LER) concept is capable of
traveling up to 200 kilometers on one charge, but when not
exploring, it is attached to the crew habitat to provide private
sleeping quarters as well as a safe haven during galactic
cosmic radiation (GCR) events. The tri-ATHLETE concept is
capable of traveling alone, but when combined with a second
tri-ATHLETE, it is capable of traversing difficult terrain
while carrying a payload as large as a habitat module.
The focus of this case is to model the surface exploration similar to what is planned under Scenario 12. As
the existing scenario is well-researched, this analysis will help validate the modeling framework rather
than explicitly evaluating feasibility. Both sortie missions to the surface locations of interest and the
build-up of outpost elements at the Lunar South Pole will be modeled. In addition, two excursions from
the outpost are modeled in detail, one short-distance excursion to the Malapert crater using two LERs
over approximately one week, and one long-distance excursion to the Schrodinger Crater using two
ATHLETEs over approximately 60 days.
Model Inputs
As the transportation system is well-defined for the Constellation program, the analysis scope will be
limited to the lunar system utilizing flight edges for all transports between the lunar surface locations and
low-lunar polar orbit (LLPO). The primary landing site on the lunar surface is at the Lunar South Pole
(LSP), though sortie missions access other surface sites listed in Table 30.
Figure 54: Lunar surface exploration
network. Visualization of flight edges to
various surface locations and surface edges
between excursion sites near LSP.
86
Table 30: Lunar outpost nodes.
Abbrev. Description Parameters
LLPO Low Lunar Polar Orbit 100 km x 100 km, 90°
LSP Lunar South Pole (Shackleton Crater) 89.9° S, 0° E
MC Malapert Crater 85.9° S, 12.9° E
SB Schrodinger Basin 75° S, 132.4° E
TC Tsiolkovskiy Crater 20° S, 129° E
AC Alphonsus Crater 13° S, 2.8° E
MH Marius Hills 14° N, 56° W
NB Nectaris Basin (Mare Nectaris) 15.2° S, 35.5° E
This case includes two surface edges to model excursions between LSP and Malapert Crater (MC) and
Schrodinger Basin (SB). Other edges model three variations of the descent module (DM) vehicle, a
“sortie mode” with a dedicated airlock, an “outpost mode” with no airlock, and a “cargo mode” which has
neither an airlock nor an ascent stage.
Table 31: Lunar outpost edges.
Name Origin Destination Length Capacity
Lunar Descent (Sortie) LLPO LSP, TC, AC,
MH, NB, OH
1 day 4 crew, 710 kg*
Lunar Descent (Outpost) LLPO LSP 1 day 4 crew, 1000 kg
Lunar Descent (Cargo) LLPO LSP 1 day 0 crew, 14600 kg
Lunar Ascent to LLPO LSP, TC, AC,
MH, NB, OH
LLPO 1 day 4 crew, 100 kg
Malapert Excursion LSP MC 180 km -
Schrodinger Excursion LSP SB 500 km -
* 500 kilograms baseline + 210 kilograms to support 7-day sortie exploration
Element models focus on the major surface infrastructure elements rather than the launch vehicles and
transfer spacecraft because of the usage of abstracted flight edges. The elements of focus include the
descent and ascent modules, surface rovers, habitats, and logistics carriers. The descent and ascent
modules closely follow the respective flight edges. The sortie descent module includes 210 kilograms of
integrated resources to supply a 7-day exploration, modeled as a separate zero-mass sortie consumables
resource container added to its contents.
Two ISRU plants, delivered in missions 7 and 15, generate oxygen (generic COS 203) at a rate of 1000
kilograms per year. The production demands are modeled with a linear demand model using a negative
87
rate (-2.75 kilograms per day). The ISRU plants are activated by the next crew upon delivery and are
assumed to operate continuously during crewed and un-crewed periods. All produced oxygen is stored
within the ISRU plant until it is demanded by the mission-level crew consumables models.
Table 32: Lunar outpost primary elements.
Name Dry Mass
[kg]
Max
Crew
Max Cargo
[kg] Description
SDM 13000 0 710* Descent Module (Sortie Mode)
SCC 0 - 210 Sortie Consumables Container
CDM 12000 0 14600 Descent Module (Cargo Mode)
ODM 12000 0 1000 Descent Module (Outpost Mode)
AM 3000 4 100 Ascent Module
CUR 230 2 0 Capable Unpressurized Rover
LER 4000 4 1000 Lunar Electric Rover
PUP 650 - - Portable Utility Pallet
ISRU 275 - - ISRU Plant and Tools
ATH 1200 0 10000 Tri-ATHLETE
PSU 2800 - - Power and Support Unit
PEM 6000 4 10000 Pressurized Excursion Module
PCM 7800 4 10000 Pressurized Core Module
PLM 3400 0 17500 Pressurized Logistics Module
FSPS 9500 - - Fission Surface Power System
* 500 kilograms baseline + 210 kilograms to support 7-day sortie exploration
Sparing-by-mass demand models are used to model parts and maintenance demands for the major surface
infrastructure elements, including the CURs, LERs, PUPs, ISRU plants, Tri-ATHLETEs, PSUs, PEM,
PCM, and FSPS. A default rate of 10% mass per year is used during crewed periods, and a 5% mass per
year is used during un-crewed periods. Crew members use linear demand models to generate demands for
consumables. Average daily surface demands include 2 kilograms of food, 3.5 kilograms of water, 1
kilogram of gases, 0.5 kilogram of hygiene items, and 0.5 kilogram of waste disposal items per crew
member. After delivery of the surface habitat in mission 13, the demands for water are decreased to 0.5
kilograms per person per day to account for greater water recovery rates. Demands during ascent, descent,
and in-space periods are omitted. Linear mission demand models generate demands for science payload
generally, uniquely specified for each mission. Unlike other resources, science payload demands are
assumed to include any required packaging mass.
88
Secondary element models are simplified to only mass estimates without significant surface movement or
demand models. Though they only take up supply capacity in this analysis, additional analysis could
benefit from the more detailed campaign definition.
Table 33: Lunar outpost secondary elements.
Name Dry Mass
[kg]
Max
Crew
Max Cargo
[kg] Description
ALC 400 0 500 Airlock-derived Logistics Carrier
SA 50 - - Solar Array
SOD 10 - - Small Offloading Device
PCT 170 - - Portable Communications Terminal
AAMA 270 - - Active-Active Mating Adapter
CB 100 - - Chassis Blade
CA 100 - - Chassis A (Small Mobile Base)
RA 110 - - Robotic Assistant
LSMS 190 - - Lunar Outpost Manipulator System
STM 50 - - Suit Port Transfer Module
MCT 210 - - Mobility Chassis Tool Kit
BT 85 - - Battery (9-Pack)
SSU 600 - - Structural Support Unit
Missions
The mission manifest is based on concepts included in Scenario 12. Most missions deliver crew and cargo
to an outpost at LSP, though several sortie missions explore other sites. The first few missions provide
basic exploration capability with unpressurized rovers (CURs). Additional exploration capability is
delivered in missions 4 and 6, each delivering two LERs. The Tri-ATHLETE elements are delivered in
missions 9 and 12 in preparation of unloading the habitation and logistics elements in later cargo
missions.
Elements are reconfigured between states for many of the missions to highlight different operational
conditions. While in transit and upon delivery, all elements exist in a dormant state. Upon crew arrival, all
primary surface elements are reconfigured to an active state to generate increased demands. The elements
are later reconfigured to a quiescent state upon the crew departure. Both CUR surface mobility elements
are decommissioned after mission 5, which delivers more capable LER surface mobility elements. The
two ISRU and the FSPS elements are not reconfigured to a quiescent state upon crew departure and are
assumed to operate continuously.
89
There are two missions targeted for demonstrating surface transportation and excursions. Mission 7
includes an excursion to Malapert Crater (MC) using one logistics LER to preposition cargo and two
crewed LERs. Transit spans three days on both legs of the trip, and exploration runs four days at MC.
Table 34: Lunar outpost missions 1-10.
# Date Flight(s) Element(s) Description
1 5/7/2021 Descent to LSP SDM-1, AM-1, CUR-1, SA-1,
SOD-1, PCT-1
Unmanned test flight with pre-
positioning of some surface
infrastructure
2 11/7/2021
11/14/2021
Descent to LSP
Ascent to LLPO
SDM-2, AM-2, CUR-2, SA-2,
SOD-2
7-day crewed exploration
mission (4 EVAs) with 180 kg of
science payload
3 4/7/2022
4/14/2022
Descent to MH
Ascent to LLPO
SDM-3, AM-3 7-day crewed exploration
mission (4 EVAs) with 500 kg of
science payload
4 11/7/2022 Descent to LSP CDM-1, LER-1, LER-2, PUP-1,
PUP-2, AAMA-1, RA, CB, CA,
LSMS-1, STM-1, BT-1, BT-2
Cargo delivery with 820 kg of
science payload
5 2/7/2023
2/21/2023
Descent to LSP
Ascent to LLPO
ODM-1, AM-4, MCT-1 14-day crewed exploration
mission (8 EVAs) with 660 kg of
science payload
6 10/7/2023 Descent to LSP CDM-2, LER-3, LER-4, AAMA-
2, AAMA-3, AAMA-4, LSMS-2,
PUP-3, PUP-4, STM-2
Cargo delivery with 710 kg of
science payload
7 12/7/2023
12/16/2023
12/23/2023
1/4/2024
Descent to LSP
Excursion to MC
Return to LSP
Ascent to LLPO
ODM-2, AM-5, ISRU-1 28-day crewed exploration
mission (16 EVAs) with 190 kg
of science payload plus 4-day
Malapert excursion
8 3/7/2024 Descent to NB
Ascent to LLPO
SDM-4, AM-6 7-day crewed exploration
mission (4 EVAs) with 500 kg of
science payload
9 10/7/2024 Descent to LSP CDM-3, ATH-1, ATH-2, PSU-1,
AAMA-5, BT-3, BT-4, ALC-1
Cargo delivery with 1,800 kg of
science payload
10 11/7/2024
12/3/2024
Descent to LSP
Ascent to LLPO
ODM-3, AM-7, PCT-2 28-day crewed exploration
mission (16 EVAs) with 320 kg
of science payload
Continuous human presence is achieved by mission 18. Mission 20 includes an excursion to Schrodinger
Basin (SB) using a “Lunabago” concept, in which two ATHLETE systems carrying the pressurized
excursion module (PEM) and a pressurized logistics module (PLM) travel with the crew in two LERs.
Surface transport takes 25 days to get to SB, exploration lasts for 14 days, and 45 days are provided for
the return surface transport to the outpost.
90
Table 35: Lunar outpost missions 11-21.
# Date Flight(s) Element(s) Description
11 3/7/2025 Descent to TC
Ascent to LLPO
SDM-5, AM-8 7-day crewed exploration
mission (4 EVAs) with 500 kg
of science payload
12 10/7/2025 Descent to LSP CDM-4, ATH-3, PEM, PSU-2 Cargo delivery with 60 kg of
science payload
13 11/7/2025
12/28/2025
Descent to LSP
Ascent to LLPO
ODM-4, AM-9 50-day crewed exploration
mission (4 EVAs/week) with
420 kg of science payload
14 9/7/2026 Descent to AC
Ascent to LLPO
SDM-6, AM-10 7-day crewed exploration
mission (4 EVAs) with 500 kg
of science payload
15 10/7/2026 Descent to LSP CDM-5, ATH-4, PCM, PSU-3,
ISRU-2
Cargo delivery with 0 kg of
science payload
16 12/7/2026
3/28/2027
Descent to LSP
Ascent to LLPO
ODM-5, AM-11 110-day crewed exploration
mission (4 EVAs/week) with
130 kg of science payload
17 2/7/2027 Descent to LSP CDM-6, AAMA-6, PLM-1, SSU-1 Cargo delivery with 780 kg of
science payload
18 7/7/2027
¼/2028
Descent to LSP
Ascent to LLPO
ODM-6, AM-12 180-day crewed exploration
mission (4 EVAs/week) with
280 kg of science payload
19 10/7/2027 Descent to LSP CDM-7, FSPS, ALC-2 Cargo delivery with 980 kg of
science payload
20 1/1/2028
1/26/2028
3/5/2028
6/30/2028
Descent to LSP
Excursion to SC
Return to LSP
Ascent to LLPO
ODM-7, AM-13 180-day crewed exploration
mission (4 EVAs/week) with 70
kg of science payload
21 5/7/2028 Descent to LSP CDM-8, PLM-2, SSU-2 Cargo delivery with 1760 kg of
science payload
Although there would likely be a dozen or more additional cargo and crewed missions, it would closely
resemble the ISS resupply case study and are not included in this analysis.
Analysis and Discussion
As this case study uses abstracted flight edges originating from LLPO rather than modeling the in-space
transportation in detail, there are no challenges to the spatial feasibility. The process-based bat chart for
the campaign is illustrated in Figure 55.
91
Figure 55: Lunar outpost process bat chart. Green squares are element instantiations, yellow
lines are flight transports, green lines are surface transports, orange and pink squares are
element movements and reconfigurations, blue lines are surface explorations, and black squares
are elements removed from simulation.
Demands for each of the sortie missions (4, 8, 11, and 14) can be satisfied by the integrated resources in
the SDM as the exploration duration does not exceed seven days. Logistical analysis is therefore focused
at LSP starting with mission 5, the first surface exploration longer than seven days. Figure 56 shows the
feasibility at LSP given the raw and remaining capacity of landers (after infrastructure elements) and
demands for crew consumables, science payloads, and spare parts. Packing or logistics container masses
are included for demands using packing factors of 50% for water, 100% for gases, and 120% for all other
non-science pressurized items. As expected, the campaign appears logistically feasible.
Figure 56: Lunar outpost feasibility at LSP. Demands include science payloads, crew
consumables and spares with packing mass estimates included.
92
Using other visualizations, the breakdown of demand classification can be closely analyzed. Figure 57
highlights the breakdown of demands at LSP by class of supply. Crew consumables (COS 2) show the
highest demand mass, exceeding 20 tons by 2028. The second largest class of supply, spares and
maintenance (COS 4), increases rapidly as the infrastructure mass at the outpost is accumulated. Close
inspection also illustrates the change in spares rates corresponding to crewed and un-crewed periods.
Figure 57: Lunar outpost demands at LSP. COS 2: Crew Consumables, COS 3: Operational
Items, COS 4: Maintenance and Spares, COS 5: Packing Mass (Estimated), COS 6: Science
Payload, COS 7: Waste Disposal.
The demands include the effects of consumption of oxygen produced by the two ISRU plants delivered in
mission 7 and 15. Figure 58 shows the simulated results of oxygen produced by ISRU-1 and ISRU-2
available for consumption during the exploration.
a)
b)
Figure 58: Lunar outpost ISRU production and consumption. Mass of oxygen available for
consumption from (a) ISRU-1 and (b) ISRU-2.
93
The simulation results also provide several measures of effectiveness (MOEs) to quantify the exploration
campaign. There are a total of 2,500 crew surface days across all sites including sorties and excursions to
Malapert and Schrodinger craters. The total exploration capability, the dot product of crew and enabling
infrastructure (COS 6 and COS 8) is 117.5 million crew-kg-days. The exploration capability greatly
increases towards the end of the modeled campaign because of the substantial delivered infrastructure.
Other MOEs including relative exploration capability and total launch mass are not computable in this
case because launch and in-space portions of this campaign were not considered.
a)
b)
Figure 59: Lunar outpost measures of effectiveness. (a) Crew surface days across all surface
sites. (b) Exploration capability (crew-kg-days) across all surface sites.
Conclusions
This case study modeled an extended lunar surface exploration campaign based on existing architectural
studies. Modeling details include ISRU oxygen production, dynamic spares rates for crewed versus un-
crewed periods, surface transportation for excursions, and improved water recovery rates in crew habitats.
As expected with a matured design, the aggregated demands for crew consumables and spares and
maintenance show it is a logistically feasible campaign.
Additional analysis for a lunar surface exploration campaign should inspect the excursions to Malapert
Crater and Schrodinger Basin in more detail. In particular, the feasibility of transferring resources
between the outpost and remote excursion sites may not be trivial to evaluate. Multi-transport manifesting
methods are required manifest resources necessary to support the crew on long duration excursions.
94
5.1.4 Mars Exploration
Mars has been a highly sought-after destination for human spaceflight since the early days of rocketry.
The Vision for Space Exploration announced in 2004 named Mars as one of the ultimate destinations for
which precursor missions to the lunar surface would help prepare [3]. The final report of the Human
Spaceflight Review Committee also states that “a human landing followed by an extended human
presence on Mars stands prominently above all other opportunities for exploration” [18]. Although the
Constellation Program is facing cancellation by the new US and NASA administration, human missions
to Mars still hold the focus of space exploration, albeit with new technologies to provide advanced
propulsion and protection against the harsh radiation environment in transit [20].
NASA has undertaken substantial effort to design reference architectures for conceptual missions to Mars,
the most recent publication being Mars Design Reference Architecture 5.0 (DRA 5.0), published in July
2009 [21]. This design reference describes the spacecraft and missions which could be used for the first
three excursions to the surface of Mars. The Mars exploration architecture is heavily based off of lunar
concepts from the Constellation Program, including the Ares V heavy lift launch vehicle, but also
includes advanced technology concepts such as nuclear thermal rockets (NTRs) for in-space propulsion,
zero-loss cryogenic coolers for propellant transportation, and nuclear fission reactors for surface power.
Figure 60: Mars exploration network. Visualization of space edges utilizing propulsive burns to
reach potential Martian surface sites.
This case study is focused on determining the in-space propulsive feasibility to deliver surface elements
required by DRA 5.0 and identifying driving factors to manage logistics feasibility for a crew of six.
Surface operations are not modeled in detail, though there would be significant interest in future analysis
to study two-week surface excursions using pressurized rovers.
95
Model Inputs
The nodes for the case study include the Kennedy Space Center for launches, a splashdown site in the
Pacific Ocean, a low-Earth parking orbit for assembly of in-space vehicles, a reference Mars orbit for
stationary operations on orbit, and three target surface exploration sites. Although only one exploration
site, Mawrth Vallis, is used in this preliminary analysis, there would likely be at least three similar
missions to justify the large investments in new spacecraft designs and propulsion technology.
Table 36: Mars exploration nodes.
Abbrev. Description Parameters
KSC Kennedy Space Center 28.6° N, 80.6° W
PSZ Pacific Splashdown Zone 15° N, 160° W
LEO Low Earth Checkout Orbit 407 km x 407 km, 28.5°
RMO Reference Mars Orbit 250 km x 33793 km, 70°
MV Mawrth Vallis 24° N, 19° E
GC Gale Crater 4.6° S, 137.2° E
HC Holden Crater Fan 26.4° S, 34.7° W
The edges used in this case study are a mix of both space edges using propulsive burns and flight edges
providing an abstracted path between nodes. Space edges model the launches of the Ares V rockets from
KSC to LEO, as well as the in-space transfers to Mars orbit and descent to the Martian surface. Ascent
from the Martian surface is modeled as a flight to simplify modeling efforts for the Mars Ascent Vehicle
(MAV), which receives LOX propellant via ISRU production. The launch of the crew exploration vehicle
(CEV) from KSC is also modeled as a flight to provide independence from any particular launch vehicle.
Table 37: Mars exploration edges
Name Origin Destination Duration
[days]
Propulsive Burns or
Flight Capacity
KSC Launch to LEO KSC LEO 0.1 t+0: 9.8 km/s
CEV Launch to LEO (Flight) KSC LEO 0.1 6 crew, 100 kg
TMI/ MOI (Aerocapture) LEO RMO 202 t+0: 3.7 km/s
TMI,/MOI (All Propulsivee) LEO RMO 174 t+0: 4.1 km/s
t+174: 1.7 km/s
Martian Descent RMO MV 0.1 t+0.0: 0.015 km/s
t+0.1: 0.595 km/s
Martian Ascent (Flight) MV RMO 0.1 6 crew, 250 kg
Trans-Earth Injection,
Splashdown at PSZ
RMO PSZ 201 t+0: 2.25 km/s
t+200: 0.15 km/s
96
Elements modeled include the Ares V launch vehicle and the in-space vehicles and landers for Martian
exploration. Several of the Ares V launches use a modified payload fairing which serves as an aeroshell
for aerocapture and entry into the Martian atmosphere. In-space Mars transfer vehicles (MTVs) are
assembled in low-Earth orbit. Two un-crewed MTVs pre-position the cargo lander on the Martian surface
and the habitat lander in Martian orbit. The crewed MTV carries the six crew members and the Mars
transfer habitat (MTH) to rendezvous with the habitat lander before descent and surface operations.
Table 38: Mars exploration elements
Name Mass
[kg]
Max
Crew
Max Cargo
[kg]
Fuel [kg]
(Type)
Isp
[s] Description
Ares V SRBs 106,500 0 0 685,000
PBAN
26 Ares V Solid Rocket Boosters (2)
Ares V Core 173,680 0 0 1,587,000
LOX/LH2
414 Ares V Core
Ares V Interstage 9,190 - - - - Ares V Interstage Element
Ares V EDS 26,390 0 0 253,000
LOX/LH2
449 Ares V Earth Departure Stage
Ares V PLF 9,049 - - - - Ares V Payload Fairing
NTR 37,300 0 0 59400
LH2
950 Nuclear Thermal Rocket
Inline LH2 Tank
(Cargo MTV)
10,800 0 34,100 - - LH2 Tank for Cargo Mars
Transfer Vehicle
MDAV 25,780 0 5,500 10,600
LOX/LCH4
369 Mars Descent / Ascent Vehicle
(Cargo Lander)
MAV 21,500 6 300 - - Mars Ascent Vehicle (Abstracted)
Aeroshell 42,900 0 0 - - Dual-use Aeroshell Shroud
SHAB 52,060 6 1,500 10,600
LOX/LCH4
369 Surface Habitat (Habitat Lander)
NTR-S 46,600 0 0 59,700
LH2
950 Nuclear Thermal Rocket Stage
with External Radiation Shield
Inline LH2 Tank
(Crewed MTV)
21,500 0 69,900 - - LH2 Tank for Crewed Mars
Transfer Vehicle
LST 8,900 - - - - Long Saddle Truss
LH2 Drop Tank 14,000 0 73,100 - - LH2 Drop Tank
SST 8,900 - - - - Short Saddle Truss
CFC 1,860 - 7,940 - - Contingency Food Canister
DM 1,800 - - - - Second Docking Module
MTH 27,540 6 5,300 - - Mars Transit Habitat
CEV 14,000 6 250 1000
MMH/N2O4
301 Crew Exploration Vehicle
97
As additional mass for spares is included in element mass estimates from DRA 5.0, demands primarily
originate from the crew using a linear demand model with two operational states. While in transit, the
crew demands total 7.5 kilograms per crew member daily. The approximate breakdown is 2 kilograms for
food, 3.5 kilograms for water, 1 kilogram for gases, 0.5 kilograms for hygiene items, and 0.5 kilograms
for waste disposal. While on the surface, it is assumed that the ISRU plants can provide ample oxygen to
eliminate the demand for additional gases resulting in demands of 6.5 kilograms per crew member daily.
Missions
Similar to the NEO sortie, this case study only inspects a single mission to Mars, though in practice there
would be at least three human missions planned in succession. Modeling more than one mission would
provide benefits of element reuse between sites under the assumption that the exploration sites were close
enough for automated surface transportation.
A single Mars exploration mission is divided across two transit windows approximately 26 months apart.
The first five launches of the heavy lift vehicle provide the elements to assemble the two cargo MTVs in
low-Earth orbit. The first cargo MTV contains the cargo lander and the second the habitat lander. Once
both vehicles are constructed, they depart for Mars using a trans-Mars injection burn and are aerocaptured
200 days later. The cargo lander subsequently descends to the Martian surface to commence oxygen
production while the habitat lander remains in orbit for the arrival of the crew.
The second four launches of the heavy lift vehicle provide the elements to assemble the crewed MTV in
low-Earth orbit. Once constructed, the crew of six is delivered using a human-rated launch vehicle and the
MTV departs for Mars, jettisoning an empty drop tank after the TMI burn and arriving with a Mars orbit
insertion (MOI) burn 170 days later. Once in Mars orbit, the crewed MTV docks with the habitat lander
and the crew descend to the Martian surface to perform surface operations for 530 days while the crewed
MTV remains in Mars orbit. In contingencies preventing the landing of the crew, such as a failure of the
descent/ascent vehicle, a contingency food canister (CFC) provides necessary resources for the crew in
Mars orbit until the return launch window opens.
After the exploration, the crew and surface samples use the fueled ascent vehicle to return to Mars orbit
and dock with the crewed MTV. The contingency food canister is jettisoned prior to the Earth orbit
injection (EOI) burn and the crew returns to Earth using a CEV for the final re-entry and splashdown into
the Pacific Ocean.
98
Table 39: Mars exploration mission events.
Date(s) Event Details
1/1/2035 Cargo MTVs Launch 1 Payload: NTR #1
2/1/2035 Cargo MTVs Launch 2 Payload: NTR #2
3/1/2035 Cargo MTVs Launch 3 Payload: Cargo Inline LH2 Tanks #1 and #2
4/1/2035 Cargo MTVs Launch 4 Payload: Cargo Lander (MDAV, MAV, Aeroshell)
5/1/2035 Cargo MTVs Launch 5 Payload: Habitat Lander (SHAB, Aeroshell)
6/27/2035 Cargo Lander TMI Aerocaptured on 1/15/2036
7/4/2035 Habitat Lander TMI Aerocaptured on 1/22/2036
2/6/2036 Cargo Lander Descent Landing at Mawrth Vallis
4/1/2037 Crewed MTV Launch 1 Payload: NTR-S
5/1/2037 Crewed MTV Launch 2 Payload: Inline LH2 Tank
6/1/2037 Crewed MTV Launch 3 Payload: LTS, LH2 Drop Tank
7/1/2037 Crewed MTV Launch 4 Payload: SST, CTC, DM, MTH, CEV
8/22/2037 Crew Launch (Flight) Payload: CEV (6 Crew)
8/27/2037 Crewed MTV TMI Propulsive MOI on 2/17/2038
2/19/2038 Habitat Lander Descent Landing at Mawrth Vallis
8/5/2039 MAV Ascent (Flight) Payload: MAV (6 Crew, Samples)
8/10/2039 Crewed MTV TEI Pacific Splashdown on 2/27/2040
Analysis and Discussion
The first phase of analysis focuses on determining the propulsive feasibility of the campaign. All
propulsive mission transports are modeled with the exception of the crewed transport to LEO and the
Martian ascent, both abstracted with flights. In addition, notional resource containers are used to hold the
maximum cargo capacity for each of the carrier elements to provide a reasonable replication of the
manifesting process. For example, a mass-less resource container packed with 4,500 kilograms of
consumables is created inside the MDAV.
99
Figure 61: Mars exploration process bat chart. Green squares are element instantiations, red
lines are space transports, yellow lines are flight transports, orange squares are element
movements and reconfigurations, and the blue line is a surface exploration.
All space transports in the Martian exploration are found to be propulsively feasible with baseline element
properties. The propellant margins for descent, however, are especially tight. The cargo lander has a
propellant margin of 627 kilograms (5.9%) and the habitat lander a margin of 435 kilogram (4.1%).
Since the in-space propulsion depends on advanced nuclear thermal technology, a sensitivity analysis was
performed on the specific impulse of the rocket, shown in Table 40. For a specific impulse of 850
seconds, the low end of the ranges cited in DRA 5.0, the transportation system becomes infeasible, with a
residual delta-v of 160 meters per second for the in-space burn. This result indicates the performance of
the NTR has a large impact on the performance of the existing architecture.
Table 40: Nuclear thermal rocket specific impulse sensitivity.
Element Group LH2 Fuel
Capacity [kg]
Remaining Fuel [kg]
Isp = 950 s Isp = 900 s Isp = 875 s
Cargo MTV #1
(Cargo Lander) 93,500 14,100 10,600 8,700
Cargo MTV #2
(Habitat Lander) 93,500 13,900 10,300 8,400
Crewed MTV 202,700 4,900 1,500 (infeasible)
Though the mission appears propulsively feasible, the default crew demand rates exceed the available
capacity for the transportation system. Estimated demands, shown in Figure 62, are significantly larger
than the 2,650 kilogram allowance for each of the transports to and from Mars and the 7,000 kilograms
allowance for surface operations specified in DRA 5.0.
100
Figure 62: Baseline Mars exploration crew demands. Under baseline crew member demand
rates with a crew of six, demands total 8.1 tons during transport to Mars, 20.8 tons during surface
operations, and 9.3 tons during return transport.
This is, in part, because the baseline demand rates to not take into account closed-loop environmental
controls and life support systems (ECLSS) which could significantly reduce the demands for water and
waste disposal resources. Assuming a 95% water closure rate and increased usage of reusable hygiene and
waste disposal items, crew demand rates are decreased from 7.5 to 3.375 kilograms per person per day
while in transit and from 6.5 to 2.375 kilograms per person per day while on the surface.
Figure 63: Modified Mars exploration crew demands. Using a 95% water closure loop and
reusable hygiene and waste disposal items, demands are be reduced to 3.6 tons during transport
to Mars, 7.6 tons during surface operations, and 4.1 tons during return transport.
101
The consumables capacities in DRA 5.0 cannot support the modified demand rates either during transport
or during surface operations, even without taking into account the tare mass of any resource containers
that would be needed. To close the logistics loop for consumables, design changes are suggested to the
MTH for transport and the SHAB and MDAV for surface demands.
Although the delivery of the MTH and its components to LEO is not closely-constrained with the Ares V
launch vehicle, the crewed MTV is a bottleneck of the in-space transports. As the crewed MTV cannot
support additional resources for consumption during the infeasible transport, a “creative” solution is
sought using excess capacity on the cargo MTVs.
Current designs of the MTH use a contingency food canister (CFC) to supply sufficient consumables to
sustain the crew until the TEI launch window opens if all or part of the Martian surface operations were
aborted. If a secondary CFC were carried aboard either of the cargo MTV transports for use in the case of
an emergency in Mars orbit, the first could be used to satisfy in-space transport demands. This solution
assumes that consumables could safely be stored in a CFC for up to five years, a secondary CFC could be
manifested on a cargo MTV launch, the primary CFC would be accessible during transit to Mars, and the
secondary CFC could be accessible in Mars orbit.
A sample implementation of this solution includes a secondary CFC for the cargo MTV launch 3,
currently the least mass-constrained. The secondary CFC is coupled with the habitat lander MTV for
transport and docked with the crewed MTV in orbit in advance of the surface exploration. As before, all
required resources are transferred to the MTH and both CFCs are jettisoned before the TEI transport.
To support the additional demands during surface exploration, either the MDAV or the SHAB must have
an increased capacity. If the MDAV resource capacity is increased from 5,500 to 6,500 kilograms, the
cargo lander propellant margin is reduced to 465 kilograms (4.4%). If the SHAB consumables capacity is
increased from 1,500 to 2,000 kilograms, the habitat lander propellant margin is reduced to 465 kilograms
(3.2%). Since the descent stages have narrow propellant margins for both landers, the additional capacity
may very well come at the cost of scientific and exploration equipment.
102
Figure 64: Modified Mars exploration transport capacity. An additional CFC is included to
satisfy demands during transport and the cargo capacity for the MDAV and SHAB have been
increased to satisfy demands during surface operations.
Conclusions
The Mars exploration architecture as described by DRA 5.0 is propulsively feasible as specified and
logistically feasible with the addition of aggressive water recovery and some alterations to increase supply
capacity. The most constrained transportation legs include the SHAB and MDAV descents to the Martian
surface and the crewed MTV transit to Mars. Although this analysis focused on the feasibility of a nuclear
thermal rocket propulsion option, a similar analysis could inspect a chemical propulsion option as well.
The next step of analysis could model the surface operations in much more detail. Currently, the cargo
lander and habitat lander are modeled as single elements; however they are actually comprised of many
components including pressurized, unpressurized, and robotic rovers, science equipment, stationary
power systems, and in situ resource generation plants. Modeling these elements separately could provide
much more detailed information for surface logistics, including the accumulation of ISRU resources and
the option of analyzing spare parts demands on a per-element basis. Of particular interest is a two-week
excursion in which two crew members use a pressurized rover to travel upwards of 100 kilometers at
speeds of three kilometers per second.
103
5.2 Usability Experimentation
In addition to demonstrating the modeling capabilities of the SpaceNet modeling framework with case
studies, a parallel task highlights its usability. Substantial effort was placed into the SpaceNet GUI to
provide an intuitive and efficient interface for modeling and simulating space exploration campaigns.
A user experiment was designed to quantitatively evaluate SpaceNet 2.5 usability. The primary goal of
the study is to measure the time required to model, evaluate, and resolve feasibility for a relevant space
exploration scenario. It is hypothesized that SpaceNet enables faster analysis of exploration campaigns,
even for users with little experience, over independent spreadsheet-based analysis techniques. To this end,
a mission based on the NEO Sortie case study (see Section 5.1.2) was developed for users to implement
and analyze. The scenario as specified is not feasible, so users must determine what changes are necessary
to establish propulsive and logistical feasibility. The entire scenario is designed to take 30-60 minutes to
solve and there is no single solution.
5.2.1 Testing Procedure
Seven volunteer subjects were selected to participate in the usability testing. All volunteers are students in
aerospace engineering familiar with general concepts of space systems including specific impulse, rocket
staging, and in-space transportation methods. None have previous experience using SpaceNet; though
several have previously experience in modeling spacecraft and exploration using other methods. Five of
the users serve as treatment subjects using SpaceNet to perform the analysis and two serve as a control
subjects using independent spreadsheet analysis techniques.
Table 41: User experiment subject comparison.
Treatment Subjects Control Subjects
Quantity 5 2
Selected From Astronautics students Astronautics students with
experience in spacecraft modeling
and architecture
Guided Tutorial 7-day lunar sortie mission
modeled in SpaceNet
7-day lunar sortie mission
modeled in a spreadsheet
Task Evaluate feasibility of a 14-
day NEO exploration
Evaluate feasibility of a 14-day
NEO exploration
Analysis Tools SpaceNet Spreadsheet
104
Treatment Subjects Procedure
Treatment subjects are tested one at a time using a computer using only the SpaceNet 2.5 application.
Though the goal of the experiment is to measure usability, it is not designed to measure first-time learning
or discovery. A 30-minute guided tutorial, derived from the “quick start” tutorials in user documentation,
serves as an orientation to the SpaceNet modeling framework and application. The subject walks through
a 7-day lunar surface sortie mission using Constellation-class spacecraft to learn the concepts of element
instantiation, space transports and burn sequences, demand generation, and cargo manifesting.
After the tutorial session, the subjects receive with a written summary of the test scenario including
mission objectives, element specifications and constraints for modification, and a spreadsheet database
containing initial object definitions compatible with SpaceNet. The subject must model the exploration
from scratch using the existing object definitions, evaluate feasibility, and iterate on element designs to
establish feasibility. The subject may ask for help on SpaceNet functions and spacecraft design concepts
during the trial but makes decisions and performs analysis independently. The test concludes when a
feasible solution has been found. Finally, the subject completes a post-testing survey to provide feedback
on the testing experience.
Control Subjects Procedure
Control subjects perform analysis using independent spreadsheet techniques rather than SpaceNet. As the
process to evaluate feasibility without a tool requires deeper understanding technical details, control
subjects were selected based on experience with similar space system architecture problems. Similar to
the treatment group, control subjects receive a sample spreadsheet-based analysis of the lunar sortie
tutorial used to prepare the experimental subjects as an introduction to valid techniques. Both propulsive
and logistical feasibility analyses are included to guide the required analysis for the test scenario.
During the trial, control subjects receive the same written task summary of the test scenario and a
spreadsheet database containing the initial element specifications. The control subject must model the
exploration from scratch using any methods or tools he or she deems useful within spreadsheets, evaluate
feasibility, and iterate element definitions to find a feasible design. After finding a feasible campaign, the
test subjects are asked to comment on the analysis process, including methods used, difficulties incurred.
Test Scenario Description
The goal of the scenario is to evaluate the feasibility of a 14-day exploration to near-Earth asteroid 1999-
AO10 in 2025 using two crew members and modified Constellation-class spacecraft. Demands for
resources are generated by the crew members, each requiring 7.5 kilograms of generic crew provisions
105
per day. The long-duration transports required to reach 1999-AO10 require significant resources to satisfy
crew demands, a new challenge not covered in the tutorial scenario.
a)
b)
Figure 65: User experiment baseline spacecraft. (a) Modified Ares V launch vehicle. (b)
Modified Orion crew exploration vehicle.
The initial scenario definition neither has sufficient cargo payload for manifested resources nor sufficient
propellant to achieve the trans-Earth injection burn. Modifications must be proposed to the elements to
overcome the infeasibilities. The following constraints are placed on the user‟s decisions:
1. No new elements may be instantiated
2. The launch vehicle architecture cannot be altered (i.e. burn sequence)
3. The fuel type and specific impulse may not be changed for propulsive elements
4. No element dry mass may be reduced by more than 10% from the baseline
5. No fuel amount may be increased by more than 20% from the baseline
Users are instructed to use engineering judgment to limit the number of modifications required for
feasibility, also considering providing adequate propellant margins for contingencies and retaining
science payload mass for performing exploration.
5.2.2 Results and Discussion
The results for the experiment are divided into the test subjects and the control subjects, followed by a
comparison discussion.
106
Treatment Subject Results
All five SpaceNet subjects found a feasible campaign in less than 45 minutes. The total task times were
37, 35, 41, 32, and 32 minutes, resulting in a median task time of 35 minutes. In addition to the time
required to find a feasible campaign, the time to complete all event definitions and the time of propulsive
feasibility were logged, summarized with a box plot in Figure 66. In general, users first solved the
propulsive feasibility problem before investigating the logistical feasibility problem. Some users iterated
between the two problems due to insufficient propellant margin to accommodate required resources.
Each user had a slightly different sequence to find a feasible solution. For example, user 3 decreased all
available element masses before increasing fuel capacity. User 1 did not modify the launch vehicles while
all others made at least one change to the SRBs or Core elements. Two users, 2 and 5, utilized the EDS
for the first burn of the trans-Earth Injection (TEI) space transport. Table 42 lists the remaining propellant
margins and Figure 66 details the final feasible campaign designs developed by each user.
Table 42: Propulsive fuel margins.
User EDS Fuel Margin
[kg]
SM Fuel Margin
[kg]
1 479 379
2 - 796
3 1814 504
4 1696 495
5 - 2156
In general, user feedback after the experiment was positive. Many users were enthusiastic that little prior
knowledge of space exploration systems was required to perform a basic analysis. Table 43 summarizes
the post-test questionnaire providing insight to the background and experiences of each user.
107
Table 43: Post-test questionnaire results.
Ratings: 1: Strongly Disagree, 2: Disagree,
3: Neutral, 4: Agree, 5: Strongly Agree User 1 User 2 User 3 User 4 User 5 Mean
Familiar with Constellation spacecraft
architecture 3 4 4 5 5 4.2
Familiar with space exploration
modeling/analysis 4 4 5 5 4 4.4
Challenged to find a feasible campaign 3 4 4 4 3 3.6
SpaceNet identified propulsive and logical
infeasibilities 5 5 3 5 5 4.6
SpaceNet helped construct a feasible
campaign 5 5 5 4 5 4.8
Could have used another tool to perform
equivalent analysis 2 2 4 2 2 2.4
Would use SpaceNet in future analyses 5 4 3 4 5 4.2
,
a)
b)
Figure 66: User experiment results. (a) SpaceNet subject task milestones. (b) Comparison of treatment (SpaceNet) and control
(spreadsheet) subjects’ total task time. Whiskers show minimum and maximum recorded values.
Table 44: User experiment feasible designs.
User SRBs Core IS EDS SA SM CM LAS Crew EDS TEI
Burn Mass Fuel Mass Fuel Mass Mass Fuel Mass Mass Fuel Cargo Mass Cargo Mass Mass
1 - - - - - -10% +20% - -10% +20% - -10% +67% - - -
2 - - - +20% - - +20% - - +20% - - +67% - - Yes
3 -10% - -10% +20% -10% -10% +20% -10% -10% +20% - -10% +156% -10% -10% -
4 -10% +20% - +20% -10% - - - -10% +20% - -10% +67% -10% - -
5 -10% +20% - - - - +20% - - - - -10% +68% - - Yes
6* - - - +15% - - +15% - - +20% +80% -5% +20% - - -
7* - - - +20% - - +19% - - +14% - - +63% - - -
* Control Subjects
0
5
10
15
20
25
30
35
40
45
Events Complete Propulsively Feasible Logistically Feasible
Tim
e [m
in]
0
20
40
60
80
100
120
140
Treatment
(SpaceNet)
Control
(Spreadsheets)
Tim
e [m
in]
Control Subject Results
The two control subjects logged 127 and 98 minutes to complete the task. In both cases, the scenario was
modeled using a sequence of events in rows with formulas to track stack mass, demands, and propellant.
A substantial portion of the time (96 and 73 minutes respectively) was devoted to baseline scenario
modeling before any analysis was performed. In one of the cases, there were also a few rework loops
where an error was detected, requiring time to uncover it and correct the scenario.
The analysis process was very similar to that of the test subjects, first the scenario was modeled as a
whole, next element definitions were modified to accommodate the resource supplies and achieve the
required delta-v. The only modeling difference from the test subjects was a simplifying assumption that
the logistics containers are jettisoned before each propulsive burn.
5.2.3 Discussion
Other than the time required to perform analysis, there were no significant differences between the
treatment and the control subjects‟ feasible solutions. The treatment subjects completed the analysis and
provided a feasible solution in an average of 35 minutes, several times faster than the control subjects.
Perhaps more importantly, the test subjects modeled the initial scenario in an average of 12 minutes,
about 15% of the average time for control subjects to reach a baseline model without including errors
uncovered later. This is important for users who may not have experience with modeling exploration
campaigns, as creating custom models in spreadsheets can require significant research, modeling
considerations, and error tracking.
All experiment subjects enjoyed working through the scenario, both using SpaceNet and with independent
analysis tools. It is an important realization that the analysis required for verifying logistical feasibility of
conceptual space explorations is not seen as drudgery to be avoided. As the modeling framework is
flexible to analyze a wide range of general space exploration campaigns, a tool that does not require
significant background research or time may capture non-traditional users‟ interests, leading to more
researched campaign proposals.
110
6 Conclusions
The refinement of the space logistics modeling framework and implementation in SpaceNet provides a
novel and useful simulation and modeling tool. Care has been taken throughout the design cycle to
provide flexibility to analyze a wide range of interesting and relevant space exploration scenarios while
enabling the expansion for future capabilities through modular interfaces. Demonstrations of SpaceNet
highlight the capability to model vastly different exploration campaigns at varying levels of fidelity while
providing a user interface that is efficient and easy to use.
6.1 Significant Contributions
Many significant contributions exit in the revised modeling framework and software implementation.
Modeling improvements enable new and more detailed concepts to be represented, enabling the analysis
of new space exploration campaigns. Usability improvements help users model explorations efficiently
and with greater feedback than before. Some of the most visible improvements are listed below.
Modeling Improvements:
Arbitrary-Burn Space Transports
Surface Transports and Vehicles
Flight Transports
Multi-destination Scenarios
Element-centric Demand Models
Modular Demand Model Interface
Reconfigurable Element States
Common Spares and Scavenging
Repairable Parts, Auto-Repair
Transport-level Manifesting
Heuristic Auto-Manifesting
Modular Data Source Interface
Usability Improvements:
Completely Redesigned User Interface
Element-based Bat Chart
Scenario Feasibility Chart (Cumulative
Delivery Capability and Demands)
Time-Expanded Supply Network
Visualization
Repairability Effectiveness Chart
Continuous Pre-Simulation for Real-time
Error Feedback
Robust Simulation Error Handling
Integrated Database Editor
Element Sizing Tool
6.2 Future Work
With the advancements in the modeling framework and user interface, there are several areas that could
be targeted for future development and research.
First, there is a need for more detailed demand models to improve existing analyses. With the
establishment of the modular demand model interface, developing and integrating new demand models
111
should be much easier than previously possible. More advanced models are needed to model spare parts
demands, which currently depend on gross assumptions using sparing by mass models. Additional efforts
should be focused on modeling electricity demand and production, one of the major portions of space
exploration missing from the existing model. Electricity is different from other resources in that it is
mass-less and generated by a power system and it will require refinement of production demand models,
currently in limited use for ISRU plants.
Task analysis is another area for future development. Agents (both crew and robotic) should be capable of
performing a task, a specialized event, which effectively makes available time another resource to
manage. Tasks may include infrastructure deployment, maintenance or repair activities, or even scientific
exploration. By adding tasks to the existing space exploration campaign definition, a finer level of detail
is uncovered, leading to improved demand models and performance metrics.
As outlined in Appendix B, there is a distinct need for optimal multi-transport manifesting methods. For
all but the most trivial campaigns or campaigns with a large margin of capacity, the heuristic auto-
manifesting option only provides a starting point for manual manifesting which is a tedious and involved
process. In addition to automated manifesting methods, additional options to account for varying cargo
environments (pressurized and unpressurized), maximum allowable packing fractions, and required
fixtures and support equipment (FSE) will help to introduce some of the finer-level challenges and
inefficiencies of managing realistic cargo. There should also be options for user-defined cargo containers
to help perform relevant trades.
Finally, as one of the long-term goals of space exploration campaign analysis, there is significant value in
adding stochastic capabilities to the existing deterministic analysis. Though the existing domain modeling
framework could represent stochastic demand generation to represent part failures or stochastic event
generation to represent element failures, there is a significant body of prerequisite research as to how the
campaign modeling framework should react as uncertainties are resolved. These questions are covered in
detail in the Integrated Analysis Strategic Roadmap.
6.3 Integrated Analysis Strategic Roadmap
The modeling framework presented is a solid foundation on which to base future development, though
there are some limitations to present analysis capabilities. Two areas have been identified as significant
expansions for future development: expanding the analysis scope and the addition of stochastic analysis.
112
The scope of analysis within SpaceNet is currently confined to a single rigid campaign structure – any
mission sequencing, operations, or element design deviations from the stated campaign require time-
intensive sensitivity or trade studies. Ideally, one should be able to set the variables for complete
campaign optimization, presently limited to the manifesting process for cargo.
Second, demands and events are currently considered deterministically. Some demands, such as crew
provisions, have evolved from decades of research and can be consistently estimated with a reasonable
degree of accuracy [43]. Demands for spares and other failure effects, however, are clearly not
predictable, especially in the case of conceptual elements with unknown design parameters. Though there
has been significant research into stochastic sparing models, in particular by Kline and Bachman [26],
integration into existing campaign analysis is desired.
As a strategic roadmap forward, some thoughts on these two analysis expansions are listed in the
following sections.
6.3.1 Analysis Scope
The analysis scope can be broken down into four levels focusing on three separate sub-problems of
increasing difficulty, shown in Table 45. Analysis at level 0 simply evaluates feasibility of fully-defined
campaigns. Present analysis capability is near Level 1, which considers the problem of how to manifest
cargo to satisfy demands throughout a campaign. Level 2 analysis considers the problem of how to
schedule missions and explorations with known architectures. Level 3 analysis considers the problem of
how to architect elements and missions.
Table 45: Levels of space exploration analysis.
Analysis
Level
Manifesting
Problem
Scheduling
Problem
Architecting
Problem
0
1* X
2 X X
3 X X X
* Existing level of analysis is near Level 1
Manifesting Problem
The manifesting problem considers cargo placement a variable in the campaign definition. This level of
analysis is most useful for evaluating or optimizing schedule-constrained campaigns. The solution to the
manifesting problem is a definition of the creation and movement of resources through the time-expanded
113
network using defined elements and transport mission events. An optimal cargo manifest can be found
given campaign goals as an objective function.
Figure 67: Manifesting problem diagram. Optimization of the manifesting problem produces a
cargo manifest to support exploration goals during simulation.
The SpaceNet model includes a rudimentary ability to solve the manifesting problem, though optimal
manifesting methods have yet to be implemented within the modeling framework (See Appendix B).
Scheduling Problem
The scheduling problem considers mission definition a variable in the campaign definition. This level of
analysis is most useful for partially-constrained campaign definitions where the mission schedule is not
known in detail. An example campaign that would benefit from this level of analysis is the ISS Resupply
use case, in which the capabilities of the elements are known but the schedule for resupply is variable. An
optimal mission schedule can be found given campaign goals as an objective function.
Figure 68: Scheduling problem diagram. Optimization of the scheduling problem produces a
mission sequence and cargo manifest to support exploration goals during simulation.
The scheduling problem is the clear next step in analysis for SpaceNet. Goals for campaigns, such as the
continued human habitation of ISS or the establishment of a lunar outpost, are often known far in advance
of the mission sequence. By leaving the sequencing as a variable, the campaign can be quickly modified
to account for changes and unexpected events.
Architecting Problem
The architecting problem considers the detailed element design a variable in the campaign definition. This
level of analysis is most useful for conceptual campaign analysis used to drive element design
114
requirements. Examples of campaigns that would benefit from this level of analysis include long-term
Lunar and Martian explorations. The solution to the architecting problem is a design description of the
elements and missions to optimize the campaign value.
Figure 69: Architecting problem diagram. Optimization of the architecture problem produces
mission and element architectures, a mission sequence, and a cargo manifest to support
exploration goals during simulation.
Due to the immense design space for space exploration campaigns, the architecting problem may only be
tractable for a handful of decisions, if at all. Traditionally, trade studies are performed to inspect ranges of
design variables for elements or missions, an analysis framework tackling the architecture problem may
only serve as an interface to automated trade studies. It is important to note that the solution to the
scheduling and manifesting problems will likely change for each conceptual element design, further
complicating the solution of the architecting problem.
6.3.2 Stochastic Analysis
The move towards stochastic analysis necessitates an important distinction between a campaign and a
scenario: a scenario being an instance of a campaign taking into account a particular set of random
variables driving demand generation or event execution. In other words, a scenario is one possible
execution of a planned campaign. In deterministic analysis, the two terms are interchangeable.
This section only outlines the effects of stochastic analysis at level 1, that is, for manifest optimization.
Two components to stochastic analysis considered include stochastic demands and stochastic events.
Stochastic Demands
From the framework of the SpaceNet model, stochastic demands could be represented by demand models
that need not generate equivalent demands in two simulations. Stochastic demand models would be a
simple extension from the existing demand models, though the manifesting process would be quite
different. In the deterministic case, demands estimated by the demand simulator can be used to create a
manifest which satisfies all demands during the campaign, resulting in a campaign valuation. With
stochastic demands, however, there is no single case for which to manifest, rather, the demand simulator
115
must simulate many scenarios to build a distribution of demands. The demands must then be prioritized
and systematically selected to build a robust manifest.
a)
b)
Figure 70: Deterministic versus stochastic demands. (a) Deterministic demands provide a
single campaign valuation. (b) Stochastic demands require stochastic manifesting and result in a
distribution of campaign valuations.
Under some scenarios, the manifested demands will likely be insufficient. The unsatisfied demands alone,
however, do not have an effect on the scenario. Stochastic events provide a means to show effect from
insufficient demands.
Stochastic Events
Stochastic events represent event executions that could have unexpected outcomes. For example, if an
element is to enter a quiescent state if there are insufficient resources to satisfy a demand, its effect on the
overall campaign is uncertain until execution time of a particular scenario. Similarly, if resupply mission
failures were modeled as stochastic events, they would alter the existence and distribution of resources,
clearly affecting the forward simulation.
The challenge with stochastic events is that the forward-looking scenario may change after each event
execution, necessitating re-evaluation of any optimized quantities. As illustrated in Figure 71 for an
example level 1 analysis, the manifest must be re-evaluated after each event execution.
116
Figure 71: Stochastic event simulation. The manifesting process must be continuously revised
during simulation to react to resolved uncertainties.
In a brute force implementation where the manifest is updated after every event, this addition would drive
the simulation time from O(n) to O(n2) where n is the number of events in a campaign. This increase in
operational complexity is unlikely to be tractable for campaigns of interest, which may have hundreds or
thousands of events and many samples of each simulation for Monte Carlo analysis. Methods should be
developed to partially decouple the optimization phase (manifestor, scheduler, and architect) and
simulation to get closer to the determinate case.
117
Appendix A Object Model Diagrams
This appendix presents object model diagrams to serve as a visual companion to the modeling framework
description. Although there are several types of object model diagrams used, the ones presented here are
simplified to highlight the most important aspects of the model.
Abstract object classes are displayed as boxes with an italicized font. Abstract object classes primarily
provide a design interface to group concrete implementations in a way that is extensible to future
development. Concrete objects classes are displayed as boxes with a standard font. Concrete object
classes may implement an abstract class or may provide other functionality.
Subclass relationships, indicated with a white triangle, identify specialized object classes of the super-
class that can be substituted for any generic super-class application. Aggregation relationships, indicated
with a white diamond, identify access to related object classes that may change over the course of the
object lifetime. This differs from composition relationships, indicated with a black diamond, which
identify permanent ownership of related object classes.
In the example illustrated in Figure 72, the Vehicle abstract object class serves as an interface to the
concrete object subclasses Car and Airplane. The Car object class is a composition of Engine using the
assumption that a car‟s engine does not change and it is an aggregation of Passenger.
Figure 72: Example object model diagram. The Vehicle interface has two subclass objects, Car
and Airplane. A car is comprised of an engine object and aggregates Passenger objects.
118
Figure 73: Location objects. Locations aggregate Elements. The Location interface is expanded
by the Node and Edge interface, which both have several subclasses.
Figure 74: Resource objects. Three subclasses of the Resource interface include Discrete,
Generic, and Continuous Resources.
Figure 75: Element objects. The Element interface is comprised of States and Parts. States are
comprised of Element Demand Models and Parts aggregate Resources.
119
Figure 76: Element demand model objects. Element Demand Models aggregate Resources.
Subclasses include Linear and Sparing by Mass Demand Models.
Figure 77: Element object hierarchy. All elements inherit attributes from the base Element
class. Subclasses include Human Agents, Resource Containers, and Carriers. Resource Tanks are
subclasses of Resource Containers and Surface and Propulsive Vehicles are subclasses of
Carriers.
120
Figure 78: Campaign objects. A campaign model is comprised of a Network, Mission, and
Manifest models.
Figure 79: Mission demand model objects. Mission demand models are very similar to element
demand models in that they aggregate Resources.
Figure 80: Core event objects. Core element-based events include creation, movement, removal,
and reconfiguration. Core resource-based events include addition, demand, and transfer.
121
Figure 81: Composite event objects. Composite events use sequences of core events to build up
to complex functionality.
122
Appendix B Optimal Manifesting Methods
The manifesting problem, of determining what cargo to place on which flight to satisfy demands for
resources during exploration, is an important aspect of space exploration campaigns. A well-planned
manifesting policy with an optimal mix of pre-positioned, carry-along, and re-supplied cargo is essential
for balancing risks, ensuring robustness against delays and cancellations, and achieving maximum
possible exploration capability and overall mission success.
The existing SpaceNet manifesting model provides a limited ability to generate cargo manifests
leveraging the multi-transport capabilities of space exploration campaigns. This appendix introduces
ongoing research into matrix representations of cargo manifests that can be optimized using standard
techniques. The present state of research operates at a higher level than the SpaceNet modeling
framework where only generic cargo mass is considered and the details of packing into resource
containers and manifesting on specific carriers are abstracted to a simpler notion. The goal of future
research is to combine the methods to provide a unified manifesting model for optimization.
Previous work established matrix-based methods for modeling the manifesting process which has been
demonstrated with an analysis of the International Space Station resupply logistics [34]. It has been
shown in detail how the cargo manifests for a multi-flight, single-destination scenarios can be represented
with a square, M (manifest)-matrix. This existing model has been expanded upon using aspects of the
SpaceNet modeling framework allow for multi-transport, multi-node campaigns. The expanded
framework provides matrix-based methods for a wider range of scenarios including long-duration crewed
missions, necessary for any beyond low-Earth orbit exploration.
Modeling Cargo Manifests
Using an abstraction of the SpaceNet modeling framework for space exploration, a campaign is
represented using seven vector components, the origin node vector No, the destination node vector Nd, the
departure time vector Td, the arrival time vector Ta, the transport capacity vector C, the exploration
demands vector De, and the transport demands vector Dt, shown in Eq. (A1-A7) for a campaign of n
transports.
T
noooo nnnN ][ ,2,1, (A1)
T
ndddd nnnN ][ ,2,1, (A2)
123
T
ndddd tttT ][ ,2,1, (A3)
T
naaaa tttT ][ ,2,1, (A4)
T
ncccC ][ 21 (A5)
Tneeee dddD ][ ,2,1, (A6)
Tntttt dddD ][ ,2,1, (A7)
The process of manifesting maps cargo onto transports to satisfy the demands during the exploration,
rendering it feasible. A campaign‟s manifest is represented with three matrix components, shown in Eq.
(A8-A10) for a campaign with n transports.
The exploration utilization matrix, Ue, represents the utilization of cargo during exploration periods. The
element ue,ij is the mass of cargo brought by transport i that is consumed in exploration period j of the
campaign. When cargo is utilized, it is removed from the scope of the analysis – waste, packaging, and
accommodation mass are not considered. Cargo utilization between transports is only valid both
transports have the same destination node.
nnenene
neee
neee
e
uuu
uuu
uuu
U
,2,1,
2,22,21,
1,12,11,
(A8)
Crewed missions, especially ones with long-duration transports, are highly dependent on demands during
transport. These demands must be satisfied by the transport in which they originate, further constraining
the campaign. The transport utilization matrix, Ut, represents utilization of cargo during transports. The
element ut,ij is the mass of cargo brought by transport i that is consumed for transport j.
nntntnt
nttt
nttt
t
uuu
uuu
uuu
U
,2,1,
2,22,12,
1,21,11,
(A9)
The transfer matrix, T, represents cargo is transferred from one transport to another. The element τij is the
mass of cargo brought by transport i that is transferred to transport j. Transfer of cargo between transports
may not involve physical movement if the same vehicles are used in both transports. A transfer is valid
124
only if the destination node of transport i is the origin node of transport j and the arrival of transport i is
before the departure of transport j.
nnnn
n
n
21
22221
11211
(A10)
The manifest, μ, is the vector defined by the set of inputs that drive the movement of cargo throughout a
campaign, given by the valid elements of the matrices Ue, Ut, and T. Validity conditions enforce spatial
and temporal criteria for manifest actions, as shown in Eq. (A11).
|
|
|
|
|
},,{
,,
,,
,
,,,
,,,
jdiaij
joidij
ijt
jaiaije
jdidije
te
tt
nn
jiu
ttu
nnu
UU (A11)
Campaign Feasibility and Manifest Optimization
For a manifest to be feasible, three criteria must be satisfied: capacity constraints, demand satisfaction,
and mass conservation. Capacity constraints ensure the sum of exploration utilization, transport
utilization, and transferred cargo does not exceed the capacity of each transport, as shown in Eq. (A12).
icuu i
j
ijijtije )( ,, (A12)
Demand satisfaction ensures that cargo is utilized to satisfy all exploration and transportation demands, as
shown in Eq. (A13-A14).
idu ie
j
jie ,, (A13)
idu it
j
jit ,, (A14)
Mass conservation ensures that mass is not created or destroyed for each transport, with the exception for
transports that originate at a source node, represented with the set S, where resources can be created, as
shown in Eq. (A15).
125
Sniuu io
j
ijijtijeji ,,, | 0)( (A15)
The campaign is considered feasible if a solution exists to a system of linear inequalities equalities based
on the above constraints, as shown in Eq. (A16). The A-matrices impose constraints on the manifest
vector elements. Ac imposes capacity constraints from Eq. (A12), Ade imposes exploration demand
constraints from Eq. (A13), Adt imposes transport demand constraints from Eq. (A14), and Am imposes
mass conservation constraints from Eq. (A15).
0
0
such that find
m
tdt
ede
c
A
DA
DA
CA
μ (A16)
Once a campaign is deemed feasible, its manifest may be optimized subject to objectives based on some
desired policy or strategy. Potential objective functions may seek to distribute risk, improve robustness to
schedule delays or cancellations, or maximize pre-positioned resources.
126
References 1. AIAA Space Logistics Technical Committee. Retrieved 13 May 2010 from
http://info.aiaa.org/tac/SMG/SLTC
2. “A Renewed Spirit of Discovery: The President‟s Vision for U.S. Space Exploration,” NASA, February
2004.
3. “The Vision for Space Exploration,” NASA-NP-2004-01-334-HQ, February 2004.
4. “NASA's Exploration Systems Architecture Study,” NASA-TM-2005-214062, November 2005.
5. “NASA Unveils Global Exploration Strategy and Lunar Architecture”, Press Release 06-361, Houston,
Texas, December 4, 2006.
6. Shull S., Gralla E., de Weck O., “A Modeling Framework for Interplanetary Supply Chains”, AIAA-2006-
7229, AIAA Space 2006 Conference and Exposition, San Jose, California, September 19-21, 2006.
7. Taylor C., Song M., Klabjan D., de Weck O., Simchi-Levi D., "A Mathematical Model for Interplanetary
Logistics", Logistics Spectrum, Vol. 41, No. 1, 2007, pp 23-33.
8. Armar, N., de Weck, O.L., Cohanim, B., “1 versus 1.5 versus 2 Launch Architecture”, CxP IDAC-2,
NASA, August 2, 2006.
9. Whiting, J., Sturm, E., Armar, N., de Weck, O.L., “Impact of Element Propellant Types on Lunar Payload
Delivery”, CxP IDAC-2, NASA, August 15, 2006.
10. de Weck O.L., Simchi-Levi D., Shishko R., Ahn J., Gralla E., Klabjan D., Mellein J., Shull A., Siddiqi A.,
Bairstow B, Lee G., “SpaceNet v1.3 User‟s Guide”, NASA/TP-2007-214725, January 2007.
11. Shull S., de Weck O., “Modeling and Simulation of Lunar Campaign Logistics”, AIAA-2007-6244, AIAA
Space 2007 Conference and Exposition, Long Beach, California, September 18-20, 2007.
12. Shull, S. “Integrated Modeling and Simulation of Lunar Exploration Campaign Logistics”, M.S. Thesis,
Department of Aeronautics and Astronautics, Massachusetts Institute of Technology, Cambridge,
Massachusetts. 2007.
13. Lee G., de Weck O.L., Armar N., Jordan E., Shishko R., Siddiqi A., and Whiting J., “SpaceNet: Modeling
and Simulating Space Logistics”, AIAA-2008-7747, AIAA Space 2008 Conference and Exposition, San
Diego, California, September 9-11, 2008.
14. Lee G., Benzin K., de Weck O., Armar N., Siddiqi A., Jordan E., Whiting J., “Integrated Cx Mission
Modeling,” ATA-01-1006, 2007.
15. Armar N., de Weck O., “Cargo Revenue Management for Space Logistics”, AIAA-2009-6723, AIAA
Space 2009 Conference and Exposition, Pasadena, California, September 14-17, 2009.
16. Holdren, J., Letter to Christopher Scolese, NASA, May 7, 2009.
17. “U.S. Announces Review of Human Space Flight Plans,” Press Release, Washington D.C., May 7, 2009.
18. “Seeking a Human Space Flight Program Worthy of a Great Nation,” NASA, October, 2009.
19. “Fiscal Year 2011 Budget Estimates”, NASA, February 2010.
20. “Statement by Charlie Bolden”, NASA Budget Press Conference, February 1 2010.
21. “Human Exploration of Mars Design Reference Architecture 5.0,” NASA, NASA-SP-2009-566, July 2009.
22. Cirillo W., Earle K., Goodliff K., Reeves J.D., Stromgren C., Andraschko M., Merrill R.G., “Strategic
Analysis Overview,” AIAA-2008-7778, AIAA Space 2008 Conference and Exposition, San Diego,
California, September 9-11, 2008.
23. Stromgren C., Galan R., Cirillo W., “Micro-Logistics for Human Space Exploration,” AIAA-2008-7780,
AIAA Space 2008 Conference and Exhibition, San Diego, California, September 9-11 2008.
127
24. Andraschko M., Merrill R.G., Earle K., “Logistics Modeling for Lunar Exploration Systems,” AIAA-2008-
7746, AIAA Space 2008 Conference and Exposition, San Diego, California, September 9-11, 2008.
25. Siddiqi A., de Weck O. L., “Spare Parts Requirements for Space Missions with Reconfigurability and
Commonality,” Journal of Spacecraft and Rockets, Vol. 44, No. 1, 2007.
26. Kline R. C., Bachman T. C., “Estimating Spare Parts Requirements with Commonality and Redundancy,”
Journal of Spacecraft and Rockets, Vol. 44, No. 4, 2007.
27. Shull S., Gralla E., de Weck O., Shishko R., “Future of Asset Management for Human Space Exploration:
Supply Classification and an Integrated Database”, AIAA-2006-7232, AIAA Space 2006 Conference and
Exposition, San Jose, California, September 19-21, 2006.
28. “NASA Systems Engineering Handbook,” Revision 1, NASA/SP-2007-6105, NASA, 2007.
29. Gohr A., “DokuWiki”, URL: http://www.dokuwiki.org [cited 2 February 2010].
30. The Apache Software Foundation, “Apache Subversion”, URL: http://subversion.apache.org [cited 2
February 2010].
31. Eclipse.org, “Eclipse Home”, URL: http://www.eclipse.org [cited 11 June 2010].
32. Free Software Foundation, “GNU General Public License”, Version 3, 29 June 2007, URL:
http://www.gnu.org/licenses/gpl.txt [cited 1 February 2010].
33. Oracle Corporation, “Javadoc Tool Home Page”, URL: http://java.sun.com/j2se/javadoc/ [cited 2 February
2010].
34. Siddiqi, A., de Weck, O., Lee, G., & Shull, S. “Matrix Modeling Methods for Spaceflight Campaign
Logistics Analysis”, Journal of Spacecraft and Rockets, Vol. 46, No. 5, 2009, pp. 1037-1048.
35. “NASA Awards Space Station Commercial Resupply Services Contracts,” Contract Release C08-069,
NASA, December 2008.
36. Macias B., “Cargo Resupply Analysis for the International Space Station, Post Assembly Complete
Timeframe,” AIAA-2000-0007, 38th Aerospace Sciences Meeting and Exhibit, Reno, Nevada, January 10-
13, 2000.
37. “Consolidated Launch Manifest,” NASA, URL:
http://www.nasa.gov/mission_pages/station/structure/iss_manifest.html [cited 24 June 2010].
38. Smith E. A., “A Manned Flyby Mission to Eros,” Proceedings of the Third Space Congress, “The
Challenge of Space,” Cocoa Beach, Florida, March 7-10, 1966, pp 137-155.
39. Landis R., Korsmeyer D., Abell P., Adamo D., “A Piloted Orion Flight Mission to a Near-Earth Object: A
Feasibility Study,” AIAA-2007-6168, AIAA Space 2007 Conference and Exposition, Long Beach,
California, September 18-20, 2007.
40. Abell P. A., Korsmeyer D. J., Landis R. R., Jones T. D., Adamo D. R., Morrison D. D., Lemke L. G.,
Gonzales A. A., Gershman R., Sweetser T. H., Johnson L. L., Lu E., “Scientific exploration of near-Earth
objects via the Orion Crew Exploration Vehicle,” Meteoritics & Planetary Science, Vol. 44, No. 12, 2009,
pp. 1825-1836.
41. Mazanek D. D., Troutman P. A., Culbert C. J., Leonard M. J., Spexarth G., “Surface Buildup Scenarios and
Outpost Architectures for Lunar Exploration,” IEEE-1093, 2009 IEEE Aerospace Conference, Big Sky,
Montana, March 7-14, 2009.
42. Kennedy K. J., Toups L. D., Rudisill M., “Constellation Architecture Team – Lunar Scenario 12.0
Habitation Overview,” JSC-CN-19362, Earth and Space 2010 Conference, Honolulu, Hawaii, March 15-
17, 2010.
43. Larson W., Pranke L., “Human Spaceflight: Mission Analysis and Design”, McGraw-Hill, 1999.